Google Personalized Search is a personalized search feature of Google Search, which was introduced in 2004. All searches on Google Search are associated with the recording of a browser cookie. Then, when a user searches, search results are not only based on the relevance of each web page with search terms, but also on websites where users (or others using the same browser) are visited through the previous search results. This provides a more personal experience that can improve the relevance of search results for specific users. Such filtering may also have some side effects, such as creating a filter bubble.
Changes in Google's search algorithms in recent years are less concerned with user data, which means that the impact of personalized search is limited to search results. Acting on criticism, Google also allows to turn off the feature.
Video Google Personalized Search
Histori
Personalized Search was originally introduced on 29 March 2004 as a beta test of the Google Labs project. On April 20, 2005, it was made available as a non-beta service, but is still separate from regular Google Search. On November 11, 2005, it became part of the normal Google Search, but only for users with Google Accounts.
Beginning on December 4, 2009, Personalized Search is applied to all Google Search users, including those who are not signed in to your Google Account.
In addition to tailoring results based on personal behaviors and interests associated with Google Accounts, Google also implemented social search results in October 2009 based on known people. Operating on the assumption that a person's peers have the same interests, these results will give a rank boost to the sites from within the "Social Circle" of the user. Both services were integrated into regular results in February 2011 and expanded results by including content shared with known users via social networks.
Maps Google Personalized Search
Data collection
Google's search algorithm is encouraged by collecting and storing web history in its data base. For unauthenticated users, Google will see browser cookies that are stored anonymously in the user's browser and compares unique strings to those stored in Google's database. Google Accounts that are signed in to Google Chrome using a user's web history to learn what sites and content you like and base on the search results presented to them. Using data provided by Google users creates profiles including gender, age, language, and interests based on past behavior using Google services.
When a user searches using Google, a keyword or term is used to generate ranking results based on the PageRank algorithm. The algorithm, according to Google, is "the link counting system and determines which pages are most important based on them.This score is then used along with many other things to determine whether a page will rank well in search." "PageRank relies on the unique and democratic nature of the web by using its broad link structure as an indicator of individual page values. In essence, Google interprets links from page A to page B as sound, according to page A, for page B However, Google sees more than just the volume of votes or links the page receives, for example, also analyzes the pages that are voting in. The sound cast by the pages that are "important" at the expense of more and helps to make other pages important. 'Using this and other factors, Google provides his views on the relative importance of the page ','
Since the search division launched the first version with a special search result in 2005 and began considering the sites visited before, new factors have been added to filter search results. According to Google, the conclusion they made after years of testing, the best indicator for deciding which results are relevant to users is the search phrase itself - not user data - and personalization of the search results is not as big as it used to be.
Harvard law professor Jonathan Zittrain debates the extent to which personalized filters distort Google's search results, saying that "the personalization effect of search has been mild". Furthermore, Google provides the ability for users to turn off personalization features if they choose, by deleting Google records about their search history and managing Google not to remember their search keywords and visiting links in the future.
Types of data collected
There are 50 factors (called 'signals' by Google) used to determine search results. The top factor in personalizing search results is:
- Locations
- Search History
- Web History
- Social Network
Each of these variables will be the personalized factor of the user's search results in the hope of quickly providing the most relevant results to the user to answer any questions asked.
Location data
Location data allows Google to provide information based on the current location and places users have visited in the past, based on GPS location from an Android smartphone or a user's IP address. Google uses this location data to provide local listings that are grouped with search results using the Google Local platform featuring detailed reviews and ratings from Zagat.
Search history
Search history was first used to personalize search results in 2005 based on previous searches and links clicked by each end user. Then, in 2009, Google announced that personalized search no longer requires users to sign in, and instead Google will use anonymous cookies in web browsers to customize search results for those who are not logged in.
Web history
Web history is different from search history, as it is a record of the actual pages users visit, but still provides factors that contribute to search result rankings. Lastly, Google data is used in search results because Google provides many demographics about users of this information, such as age, gender, location, job history, interests, and social connections.
Social networks
Google's social networking service, Google also collects these demographic data including age, gender, location, career, and friends. This plays a major role when presenting reviews and ratings of people in user circles.
Search personalization effectiveness
To determine the actual impact of search customization on end users, the researchers at Northeastern University determined in a study with users who logged in vs the control group that 11.7% of the results showed differences due to personalization. The results show that these results vary greatly according to search queries and yield ranking positions.
In the following example, Team Portent does a search query for 'JavaScript' (shown on the right) and then searches for 'Programming Textbooks' and' Books on HTML 'before searching' JavaScript, which converts search results by bringing three book lists that are not part of the original set of results. This study shows that of the various factors tested, the two with the most measurable impact are whether users log in with Google accounts and IP addresses of search users. The same study also investigated the impact of 11.7% personalization by utilizing Amazon Mechanical Turk (AMT) (the Internet market crowdsourcing and part of Amazon Web Services) vs. the control group to determine the difference between the two. The results show that the top ranking URLs are likely to be unchanged based on personalization, and that personalization occurs most often at a lower rank than the resulting page.
Reception
Some concerns have arisen regarding the feature. This reduces the possibility of finding new information, therefore bias the search results against what the user has found. It also introduces some privacy issues, because users may not realize that their search results are personalized to them, and that affects the search results of others using the same computer (unless they log on as different users). This feature also has a profound effect on the search engine optimization industry (SEO), since search results are not ranked the same way for each user - thus making it more difficult to identify the effects of SEO efforts. Personalization makes the search experience inconsistent for different users who need the SEO industry to be aware of personalized search results and not personalized to get improved rankings.
Personalized search suffers creating an abundance of background noise for search results. This can be seen as a carry-over effect in which a single search is performed followed by a subsequent search. The second search is affected by the first search if the timeout period is not set at a sufficiently high threshold. An example of the negative effect of the carry-over effect is that shop searches in Hawaii can carry previous search results, failing to show the same store in California, creating noise.
However, in recent years new research has stated that search engines do not create the kind of previously estimated filter bubbles. In a study of the political impact of search engines in seven countries conducted at Michigan State University, researchers found that search engines were complementary to other news sources that people have used. Users check an average of 4.5 news sources across various media to gain understanding, and those with a special interest in politics check for more. The researchers noted that the filter bubbles sound like real problems and that they primarily appear to apply to people other than yourself. Their conclusion is, however, that the issue is over-exaggerated, anecdotal evidence, and it is impossible to see that search engines contribute to the creation of filter bubbles based on empirical evidence generated by research.
See also
- Bubble filters
References
Source of the article : Wikipedia
0 Comments