PDF Publication Title:
Text from PDF Page: 020
Additionally, exactly because specific groups or users are targeted it is difficult to detect and regulate personalisation (Epstein and Robertson 2015, 4519). Epstein and Robertson also note that while campaign influence is usually explicit, search engine manipulations are hard to detect (Epstein and Robertson 2015, 4518). 1.2.3 Benchmarks / standards Where does personalisation ends and discrimination starts? There is no clear standard for this, although there are cases in the US that have been considered as discrimination (Kirchner 2015). Also Google itself has apparently set a line in what is considered acceptable and inacceptable: for instance, there are no auto-complete suggestions for queries that are about pornography or illegal downloading (2015), and some type of images, such as a “racist photoshopped image of Michelle Obama”, are not included in image search (Gillespie 2014, 180). While many researchers claim that algorithms, or its results, are biased or unfair, most scholars, such as Epstein and Robertson7, neglect the fact that no benchmarks exist, which makes it problematic to consider something as unfair or biased. Other scholars do acknowledge the problem of a lack of standards of what is considered fair and unfair, neutral, bias, discrimination and so forth. Jeremy Kun, for instance, raises the question: “what does it mean for an algorithm to discriminate” and concludes that there is no benchmark for what is considered discrimination and what not (Kun 2015). Dwork et al. (2011) have attempted to create a tool and algorithms “that guarantee fairness” (Miller 2015). However, in an interview with NY Times’ Caroline Miller, Dwork stresses that she approaches fairness from a mathematical point of view: “Fairness means that similar people are treated similarly . . . It would require serious thought about who should be treated similarly to whom,” she adds (Miller 2015), referring to mathematical classification schemes. She argues that ethicists should determine “whose responsibility it is” to “ensure that algorithms or software are not discriminatory” (Miller 2015). 7 Consider the following hypothetical example in relation with Epstein and Robertson’s research: candidate A and candidate B are participating in elections. The PR-person of B has accomplished that lots of media write about candidate B. This would influence search results, because in a specific period, more media have discussed person B than person A. How should a search engine behave? Should it list more results about person B (because recently more news has been about B?), or should it divide the amount of search websites equally, since there are two candidates? And if one of the candidates has been discussed in the news because of a personal incident that has affected many people’s attitude towards him or her? Should a search engine distinguish between relevance, or not? Is a search engine biased if it does show results about personal matters, or should it stick to news about politics? And if Americans tend to be more influenced by personal contextual information about candidates (think about the Monica Lewinsky affaire), but Dutch people restrict themselves to the political functioning of a person. How should a search engine decide on relevancy? 20PDF Image | peek into the discursive construction of the Google Search Algorithm: A critical discourse analysis
PDF Search Title:
peek into the discursive construction of the Google Search Algorithm: A critical discourse analysisOriginal File Name Searched:
thesis-google-search-algotithm.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com (Standard Web Page)