How Does Google And Facebook Algorithm Works And How Raters Help Them Improve

Share the love

Whether it be job hunting or fixing a bulb, Google has the answer. But did you ever wonder how does it do it? How does it work? Are there any people who check Google’s work?

Let us find the answers to these questions.

WHAT IS A SEARCH ENGINE?

A search engine, as the name suggests, is a program for searching content over the Internet. When a user types a query into the search box, the search engine searches for it in its database and provides the most relevant information to the user.

The search engine stores and arranges data from all over the World Wide Web to fetch the desired results to its searcher.

HOW DO SEARCH ENGINES WORK?

We can briefly divide the working of Search Engine into 3 basic steps-

·  Crawling– Search engines have a software called web crawler/spider that goes through content all over the web, literally and figuratively crawling from one website to another.

·  Indexing – All the data found during crawling is stored and organized according to keywords and other parameters. (Imagine this like the Index provided at the end of a big book)

·  Ranking– Each web page is ranked in the order of relevancy to a given query and then presented to you, as search results. These ranks are based on certain algorithms.

WHAT ARE ALGORITHMS?

Algorithms are the backbones of a search result. They are a combination of processes and sets of rules given to a machine to retrieve stored information systematically. Search engines keep improving and updating their algorithms to meet the needs of their users. Big companies like Google and Bing make minor changes every day, often unnoticed by a common user. Some updates are huge, often called broad core updates, for example, “Penguin” which was released by Google to catch link spam.

GOOGLE’S SEARCH ALGORITHM

Google religiously works on improving its algorithm, eliminating spams, and providing relevant search results, making it the most popular search engine in the world. Google uses PageRank as the machine learning algorithm to rank the pages in their SERP (search engine results page). Their algorithms are very complex and are based on several parameters including keywords, keyword placement, the newness of content, location, reliability of web pages, linking, establishment history, etc. The importance of these parameters depends on your queries. Google takes into consideration many factors to provide you with your prime requirement in terms of search results.

Some of these factors have been enlisted below.

1.     Interpretation of your query– Over time, Google has introduced several tools to understand the meaning and intention behind a query. Spell checker and synonym finder are two such innovative tools. Universal searches have provided you access to not only the relevant text related to your query but also images, videos, news, maps, and research papers.

2.     SERP relevance to query– The goal of these search engines is to fetch the users with relevant data. The primary consideration is matching keywords and keywords placement, whether it is in your headings or domain name or tags.

3.     Quality of content- It is often misjudged that by stuffing your webpages with keywords, you can land in the top searches. But Google’s algorithm is programmed such that it tracks spam and identifies the trusted and authentic pages.

4.     External linking of the page- A high PageRank score is often determined by the number of external links (links on other authoritative websites) that lead to your page. The higher the score is, the better ranking it deserves.

5.     Performance of a page– It also ranks webpages based on their speed, usability, and performance on different devices like smartphones, laptops etc.

6.     Established history of engagement- Rating also depends on the history of a website, as to how old and established it has been, the number of users it has attracted over a while, and the accuracy of data it provides.

To ameliorate the user experience and improve the standards of content provided by social media as well as search engines, various experiments, tests, and feedback by raters are taken into consideration.

WHAT ARE RATERS?

Raters are independent contractors hired by search engines to check the relevance and quality of the results provided in response to a search query. Their primary role is to evaluate, and rate pages based on usability, relevance, and quality. Their feedback indirectly helps search engines to improve their algorithms.

Example: You search “sale on Adidas shoes” and one of the search results shows you “history of Adidas shoes”. Surely, you’re not interested in that and it is immaterial to you in the present circumstances. A rater will help in providing the results that the searcher is interested in.

HOW RATERS HELP IN IMPROVING THE ALGORITHMS?

Search engines and social media platforms like Google and Facebook appoint people (through contractors like Lionbridge, Appen etc.) to work for them as independent agents for evaluating webpages and websites that appear as search results. These agents are called raters/web analysts/search engine evaluators/internet assessors.

One may feel that evaluating pages may be an easy job, but it is not so. These raters clear tough qualifying examinations to be appointed. They go through guidelines repeatedly to perform well and provide insightful feedback.

Raters do not directly affect the ratings of websites. Instead, their feedback is used as data for these search engines to improve their algorithms. Search engines take this data very seriously and work on ways for their betterment. The key help that a rater provides in improving the algorithm is by performing the following tasks-

URL Rating Tasks- In this type of task, raters are given a search query and a URL as the search result. They have to ‘rate’ the query on a predefined scale in order of relevance. As per the publicly available data, the most common scale involves ratings like-

  1. Vital: For special situations (e.g. official pages)
  2. Useful: For a page that is helpful for most users.
  3. Relevant: For a page that is helpful for some users.
  4. Slightly Relevant: For a page that is not very useful but is related to the query. Few users may find it relevant.
  5. Off-topic or Useless: For a page that is not relevant/helpful. Few users may find it useful.
  6. Unrelatable: For a page providing content unrelated to the query.

Page Quality Tasks- It is a query-independent task, where there are sets of questions in front of you that judges the quality of a page.

The first set of questions is related to language. You need to mark if the landing page is loading or not and flag the page for porn if any.

The second set of questions is related to the landing page. You need to   answer on the following parameters

  •      Purpose of the page
  •       Main Content
  •       Supplementary Content
  •       Layout of the page

The third set of questions is based on the website and homepage. Answers need to be marked on the following parameters

  •    Purpose of the website
  •    Responsibility of content
  • Contact Information
  •    Reputation of website
  •    Maintenance of website

Finally, you need to give an overall rating to the page varying from a lowest to highest scale and leave a necessary comment/feedback.
A rater must consider many factors while assessing the page like the purpose of the page, Expertise, Authoritativeness, and Trustworthiness (E-A-T), website reputation etc.

Needs Met Tasks– Over the years, the number of mobile phone users has tremendously increased. Therefore, Google makes sure that all their needs are met. The ratings for a resultant page depending on the query, are given below.

  •       Fully Meets- When the resultant page immediately meets the needs of the mobile users. Users may not refer to any additional page.
  •       Highly Meets- When the resultant page is very helpful. Few users may refer to additional pages.
  •       Moderately Meets- When the resultant page is helpful. Some users may refer to additional pages.
  •       Slightly Meets- When resultant pages have some relevance to the query but not enough. Most users will refer to additional pages.
  •       Fails to meet- When the resultant page fails to meet the need of the query. Most/All users will refer to additional pages.

Side by Side Tasks- The screen is split into two SERPs (containing one or many pages) of a query, one shown with an algorithm change and the other with the old algorithm. The rater is asked to choose a side depending on relevancy and rate it on a scale that tells if one side is much better than the other or they may be equally preferable. The rater must explain his reasons for choosing a given side.

Experimental Tasks- This category has new and different tasks. The set of instructions that is provided needs to be considered seriously and tasks should be performed accordingly.

Design Tasks- In this type of task, a rater assesses the design and layout of a website/webpage to check if the page provides easy access to its main contents, supplementary contents and is according to the user’s intent. More than the beauty of the page, the presentation of content is important. Deceptively designed pages are those who use their Ads as the main content and fool the users to click on monetized links or others.

Other Tasks- Here are a few examples of some other tasks performed by raters. These may/may not be included under experimental tasks.

  • They may be asked to rate the automated voice sounds depending on their resemblance to a human voice.
  • They may be asked to rate the utility of suggestions made by Google, such as autosuggestion, “Did you mean?” etc.
  • They may be asked to rate a transcribed audio file.
  • They may be asked for feedback on Google’s personalization services.
  • They may be asked for feedback on apps used by mobile users.

It is important to bring to your attention that raters do not just evaluate webpages for search engines, they also work for social media websites. Daily news, videos, ads, and content that you see on your Facebook news feeds are also assessed by raters. Let us know more about it.

FACEBOOK ALGORITHM

Facebook also uses a set of algorithms to determine which posts to show in your news feed and which order. The Facebook algorithm is very user-centric and is based a lot on the user’s interests. Facebook strives to improve user experience and constantly updates its algorithm. Below mentioned are a few major changes recently made by Facebook in its algorithm.

·       A major change that took place in its algorithm was in 2018 when the company decided to restrain from fake news and draw up posts from friends and family in the user’s newsfeed rather than that of brands.

·       Another blow to the world was its updated algorithm in 2019 when the company decided to prioritize high-quality news from expert, authoritative, and trustworthy sources that are not only relevant to the users but are also informative. Video rankings based on loyalty and intent, more personalized feed, and less spam were other major changes.

ROLE OF RATERS
While working with Facebook as a rater, you can be asked to perform one or many tasks, such as these:

·       There are openings for Ads Assessor. You will be asked to rate ads based on their quality and relevancy. You must review the Ads content, text, images, videos, etc. The ads can also be flagged for inappropriate content. You must grade them based on a rating scale and give your valuable feedback on the same.

·       Facebook also employs people (through Lionbridge/Appen) to review the videos surfacing on the website. Under Appen’s Uolo project, raters will be asked to check the credibility of the video, as to assess if the claims made by the person/company who posted the video is true or not.

·       Under Appen Connect Slickrock Project, you will be evaluating the personalized recommended stories on Facebook, for relevancy and user’s interest.

·       Other projects may include evaluation of search results, check upcoming events, etc.

IN CONCLUSION

“If search engines were a body, algorithms would be their soul.” 

Algorithms are indeed the foundation for a SERP. For its foundation to be stronger and better, search engines appoint raters. Although raters have no direct influence on the ranking of a page, their critical feedback plays a significant role in helping search engines and social media websites to upgrade the quality and utility of their content as well as to maintain their high standards.

Being A rater Is a highly paying Job Earning Up to 10 USD/ Hour Read The articles below.

Work From Home Earning Opportunities At Appen

Work From Home Opportunities At Lionbridge

Online Earning Opportunities At Teemwork.Ai

Lionbridge MAPS ANALYST Project And Appen SHASTA PROJECT

How To Get A Search Engine Evaluator Job

Lionbridge Internet Assessor And Appen Yukon Projects

Tags: , , , , , , , , , , , , , , , , , , ,

Leave Comment

Your email address will not be published. Required fields are marked *