草庐IT

Will 'Recommendations' Meet the Same Fate as 'Search'?

千山 2023-03-29 原文

Recently, Zuckerberg has had a difficult time.

The corporation has been criticized for data and privacy breaches, creating global public concern.

Even after adopting a new brand name, Meta, the social media behemoth that intended to present a new image, has not been able to resolve its dispute with the European Commission over the transfer of user data.

According to the EU's most recent data protection regulation, corporations who gather user data within the EU must keep and handle the data in this region. However, a portion of Facebook's and Instagram's data is still stored on U.S. servers, resulting in inevitable disputes between the two sides.


The Trend: Tightening Online Privacy Policies Worldwide

With big data technology, exposure to personal privacy brings a considerable bonus for online marketing. Platforms take advantage of massive amounts of data, including personal information, preferences, and user behaviors, with the intention of classifying and tracking users more accurately and comprehensively.

Therefore, companies like Meta that rely on personalized recommendation algorithms to generate revenue struggle to make ends meet under solid personal privacy protection.

In 2021, Apple rolled out significant additional privacy protections for iPhones and iPads, significantly impacting the tech industry. Under the new rules, advertisers can no longer track the effectiveness and delivery of social ads as they did before, creating big trouble for companies like Facebook and Instagram that make their money through targeted advertising.  

Since advertising accounts for 97.5 percent of Meta's revenue, the social giant's soft spot seems to be hit by Apple, as evidenced by Meta's disappointing earnings report released in early February this year.

Meanwhile, global governments have been making significant efforts to strengthen personal data regulation in light of the increasing incidence of data leaks. Living in a digital world, we all know how important data is. Not only is it necessary for a business to customize its products for its clients, but it may also reflect the economic and social position of a country. In the course of modernization, a nation may be exposed to unforeseeable threats in the absence of a data protection infrastructure.

As early as May 2018, the General Data Protection Regulation (GDPR) in the EU hindered the unlimited use of data by large Internet companies like Google and Meta. Since implementing Apple's Application Tracking Transparency feature in April 2021, many of them are losing their edges in user growth, advertising revenue, and other traditional profit-making sectors.  

The same is true for the domestic market in China. The Chinese Advertising Association, and internet titans including Tencent and Bytedance, sought to develop a mechanism dubbed China Advertising ID (CAID) to avoid Apple's privacy adjustments. The mobile device manufacturer then issued a warning that it would remove programs from the Apple Store if they circumvented the privacy feature.

Chinese Internet companies will somehow utilize data to adjust their products just like their foreign counterparts. Still, the business model, which works by developers collecting data to draw user-profiles and do advertising, seems to have reached a tipping point. With IOS devices taking the lead, it is only a matter of time before we see Android and Windows follow up to launch their comprehensive privacy protection mechanisms.


The Fall of Personalized Recommendation

The emergence of more privacy policies is caused by increasing concerns from the public, as most recommendations are built on massive user data collection, and users are often unaware of it. Think of how often you scroll down to the page and have to click "agree" without reading the terms and conditions and how accurate your preferences and personality could be portrayed by analyzing digital footprints.

Imagine having a casual chat with your friend during lunch and speaking about a fancy wristwatch because you saw a stranger who passed your table wearing it. You had never heard of this watch model before, nor did you search for it on any e-commerce platforms. However, this exact watch just popped up on the front page of your shopping app, indicating you may want to add it to your cart.

Some customers may like this kind of tailored recommendation function, as it saves time and enhances search efficiency. With the help of big data technology, your mobile apps may know better than you. From news feeding to video sharing, food delivery, to online shopping, your interests and preferences are pretty familiar to your apps, which will rely on your data to achieve higher user stickiness.

The developers, product managers, and algorithm engineers behind these platforms work hard to create "additive" products every day. It is not a rare situation that you plan to watch a 3-minute video but end up browsing it for three hours. Even if you purchase nothing, your time and attention are enough for these companies to make profits.  

In the documentary "The Social Dilemma", companies seem to have three key goals when they obtain data:

Engagement: to increase the use and keep you coming back;

Growth: to get one user to bring more users;

Advertising: to tailor the advertisements and then make money from them.

In this era of personalized recommendation, a human being could be parsed into a series of data by algorithms, added with various labels by greedy companies whose primary income could just come from those digits from users. With the advancement of technology, the recommended material appears to get more specific to your expectations, to the point where it may sound unsettling and lead to unforeseeable outcomes.

Previously, platforms offered content based on users' needs. Now, our requirements may be inferred from the offered information.

It becomes more difficult to determine if a person visits a new restaurant or purchases a bag out of genuine desire or because they have viewed their filter-enhanced photographs too frequently.

The strengthening of personal privacy protection means enterprises cannot obtain, save, or use user information without authorization or permission. Right now, it may sound too early to say that personalized recommendations will reach a dead end. Nonetheless, we should be aware that in a world driven by algorithms, humans utilize tools and are not exploited by tools.


Something New in Web 3.0?

We have heard a range of buzzwords in the past year, with Web 3.0 being one of them. With the development of the Internet, traditional search engines and personalized recommendations may not work as well as they used to, driving people to develop new tools that could suit the new iteration of the World Wide Web.

But what is Web 3.0, and how will search engines and personalized recommendations do their bits in that?

In Web 1.0, search engines like Google and Baidu quickly rose to prominence with their technologies. Users just make simple inquiries and collect information. The search engines could not make accurate guesses about your likes and dislikes; search results and browsing time were more impacted by the interface design or editor picks.

Social applications such as Facebook, Instagram, QQ, and WeChat ushered the public into the era of Web 2.0. Users' engagement with the Internet has soared, and they are more willing to generate content based on platforms. Acting as the core players, those tech giants have formed multiple ecosystems that monopolize data, value, and network effects. Users are content creators, but platforms control absolute ownership, management, and distribution. This leads to intensifying conflicts as the majority of profits go to the wallets of those giants.

Web 3.0 is an idea for a new iteration of the World Wide Web based on blockchain technology. Through technologies like encryption and distributed storage, users could get general protection for their data privacy. They will have content ownership to distribute in the next stage of the Internet, and the values created by them can be further distributed following their contracts. As Neal Mohan, chief product officer at YouTube, said in his outlook for 2022: "Web3 also opens up new opportunities for creators. We believe new technologies like blockchain and NFTs can allow creators to build deeper relationships with their fans. Together, they'll be able to collaborate on new projects and make money in ways not previously possible. For example, giving a verifiable way for fans to own unique videos, photos, art, and even experiences from their favorite creators could be a compelling prospect for creators and their audiences. "  

With Web 3.0's blockchain protocols, users' data will be extremely secure.  

With creators regaining custody of their data, search engines and tailored recommendations may usher in revolutionary innovations that forever alter our Internet experience.


有关Will 'Recommendations' Meet the Same Fate as 'Search'?的更多相关文章

  1. ruby-on-rails - 结合 meta_search 与 acts_as_taggable_on - 2

    我在开发的Rails3网站的一些搜索功能上遇到了一个小问题。我有一个简单的Post模型,如下所示:classPost我正在使用acts_as_taggable_on来更轻松地向我的帖子添加标签。当我有一个标记为“rails”的帖子并执行以下操作时,一切正常:@posts=Post.tagged_with("rails")问题是,我还想搜索帖子的标题。当我有一篇标题为“Helloworld”并标记为“rails”的帖子时,我希望能够通过搜索“hello”或“rails”来找到这篇帖子。因此,我希望标题列的LIKE语句与acts_as_taggable_on提供的tagged_with方法

  2. ruby-on-rails - 了解 "attribute_will_change!"方法 - 2

    我想覆盖store_accessor的getter。可以查到here.代码在这里:#Fileactiverecord/lib/active_record/store.rb,line74defstore_accessor(store_attribute,*keys)keys=keys.flatten_store_accessors_module.module_evaldokeys.eachdo|key|define_method("#{key}=")do|value|write_store_attribute(store_attribute,key,value)enddefine_met

  3. ruby-on-rails - 在没有 :total_entries to improve a lengthy query 的情况下使用 will_paginate - 2

    我有一个will_paginate的当前实现,它使用paginate_by_sql方法来构建要分页的集合。我们有一个针对total_entries的自定义查询,它非常复杂并且给我们的数据库带来了很大的负载。因此,我们想从分页中完全删除total_entries。换句话说,我们只需要一个“下一个-上一个”按钮,而不是“上一个1[2]345下一个”的典型分页显示。但我们需要了解一些事情。我们是否显示上一个链接?这当然只会发生在当前选择中显示的记录之前存在的记录我们是否显示下一个链接?如果显示集合中的最后一条记录,则不会显示此内容来自docsAqueryforcountingrowswill

  4. ruby-on-rails - 限制 will_paginate 中的页数 - 2

    因此,在使用Sphinx时,搜索限制为1000个结果。但是,如果will_paginate生成的结果分页链接超过1000个,请不要考虑这一点,并提供指向超过1000/per_page的页面的链接。设置最大页数或类似内容的明显方法是什么?干杯。 最佳答案 我认为最好将参数:total_entries提交给方法paginate:@posts=Post.paginate(:page=>params[:page],:per_page=>30,:total_entries=>1000)will_paginate将仅为显示1000个结果所需的页

  5. ruby-on-rails - Rails 新手 : Recommendations for error handling in controller - 2

    抱歉,如果问题很明显,我才刚刚开始使用Rails。我现在在几个Controller方法中有以下代码:respond_todo|format|if@project.saveformat.html{redirect_to(edit_project_url(@project),:notice=>'#{user.name}addedto#{role}.')}format.jselseformat.html{render:action=>"edit"}format.js#...endend那么问题来了,对于所有方法中的错误,最好的方法是什么?是否建议我使用save!并在rescue_action

  6. ruby-on-rails - Rails 设计 : "You need to sign in or sign up before continuing" instead of "You will receive an email with instructions.." - 2

    我已经安装了DeviseonRails4.2.0,一切似乎都在工作,我使用了以下指南:http://sourcey.com/rails-4-omniauth-using-devise-with-twitter-facebook-and-linkedin/我的设计模块是:devise:database_authenticatable,:registerable,:confirmable,:recoverable,:rememberable,:trackable,:validatable,:omniauthable唯一的问题是,如果我尝试通过转到注册页面创建一个新帐户,然后在输入我的电子邮

  7. ruby - @users 变量为空。忘记为 will_paginate 传递集合对象? - 2

    问题localhost:3000/users/不会显示我谦虚地进入,因为我是第一次尝试通过Rails教程。我在第10章,我已经花了5个小时解决这个问题。当我尝试访问localhost:3000/users/时出现错误(我相信这与factory_girl有关)解释了@users变量为空并且我忘记了为will_paginate传递一个集合对象。我目前在第10章第10.23节,每次运行时:$bundleexecrakedb:reset$bundleexecrakedb:populate$bundleexecrakedb:test:prepare我在解释时遇到错误rakeaborted!Fac

  8. ruby-on-rails - rails : Prevent will_paginate from calling #count of ActiveRelation - 2

    当我向will_paginate传递一个ActiveRelation时,它总是调用它的#count方法并访问数据库以找出项目的总数。但是这个操作需要时间,而且我已经缓存并准备好了总数。我可以将这个预先计算的计数传递给will_paginate并阻止它访问数据库吗?我尝试了:count选项,但它作为选项传递给了ActiveRecord:active_relation.paginate(page:2,per_page:100,count:total_count)谢谢!:) 最佳答案 通过:total_entries传递缓存计数解决了问题

  9. ruby-on-rails - pg_search 使用 associated_against 给出错误 "column [model_name].[associated_column_name] does not exist" - 2

    我正在尝试使用pg_search来搜索关联模型。当我运行搜索时,出现错误“PG::Error:ERROR:columnplans.namedoesnotexist”。我正在“计划”模型中运行搜索,并尝试针对“地点”与列“名称”的关联进行搜索。连接这些的has_many:through模型是多态的。不知何故,sql查询将两者结合起来并抛出错误。我已经运行了associated_against迁移(railsgpg_search:migration:associated_against),搜索了文档,并寻找其他有错误的人,但一无所获,一定是我只是忽略了一些东西。如果我只是删除plan.rb

  10. ruby-on-rails - rails : How do I autocomplete search for Name but save ID? - 2

    我用过这个视频http://railscasts.com/episodes/102-auto-complete-association-revised在我的应用程序的表单中设置自动完成搜索输入。(该视频可能仅供成员(member)使用,因此我也会发布我的代码。本质上,它会搜索数据库(名称)的一列,并在您键入时在下拉列表中自动完成。这一切都很好,但是,我想要什么表单要做的是提交与名称相关的ID,而不是名称本身。我假设没有简单的方法可以仅在View中执行此操作。任何帮助都会很棒。下面的代码,让我知道是否有任何其他代码有帮助。谢谢Controller:defgame_namegame.try

随机推荐