From a technical point of discussion: did Google search favor Hillary Clinton? -Google search, Hillary-IT information

Recently, the independent media company called SourceFed video points out that Google manipulated search results shielding on Hillary Clinton's negative keyword search , to support Hillary Clinton in the general election. This news came out, friends call "real life version of the House of cards."

In response, Google has denied it.

How SourceFed is questioned?

Is widely spread in this video, SourceFed said, when you want to query Clinton's black history, "Hillary Clinton crimes (crime Hillary)", type the first part of "Hillary Clinton CRI", the search box displays automatic keywords is:

Obviously, Google's second and third option links to a criminal reform Bill Clinton proposed in 1994, belonging to positive search results.

However, the search results based on user searches for keywords and more times more likely to be automatically associated keywords. For example, Bing search, Yahoo search, a negative association to ask for more:

In fact, the Clinton crime event itself even more in the spotlight, SourceFed queries to Google's statistics show, compared to Google's suggested keywords, "Clinton crime" higher search traffic:

Meanwhile, Clinton indicted gel event is of interest to the majority of users, but did not recommend Google keywords:

In contrast, Bing and Yahoo's results were as follows:

In addition, the Google search results of negative information about other candidates were compared with other engines, such as broken search:

(Popular challenge to the bed breaking racism)

Thus, SourceFed says, Google manipulated search results, giving special protection to Mrs Clinton's public image.

If true, search engine this immoral, violated people's right to know.

Google denies: modify keyword tips, but does not manipulate search results

Faced with this challenge, Google says:

"Google's automated Lenovo does not favor any candidates, these are not really questioned simply because of ignorance of the function mode. Automated keywords generated based on a variety of factors, including keyword popularity and so on. Our system regularly optimized, user activity will vary, so automated keywords will change accordingly. In addition, when together with contact names, our search engine does not automatically suggest to offend or slander issues.

Google's response has also attracted some more questions, comments:

That said, I just got you bite me?

Survey: accused of sweeping suspected

Yesterday, Google senior engineer Matt Cutts Twitter responded by saying that such accusations were "silly mistakes". Cutts pointed out that

SourceFed on the grounds that is to use the Google search for "Hillary Clinton in" automatic Association did not appear "Hillary Clinton indictment", enter "Hillary Clinton CRI" nor Clinton crime search suggestions. But it's worth noting is, in fact, most users in the search type is "Hillary ...", instead of the full name "Hillary Clinton ...". When you search for "Hillary in", it will be "Hillary indictment of" automatic associations.

Small test:

Meanwhile, Google also vilify bed broken, also result in the example above, enter the "Donald Trump RAC", as previously, just type "Donald Trump RA", in accordance with the search results display, Google also supports breaking:

Because the user "trump divorce during the rape (rape) his wife" a concern than "Rally (rally)" high.

Moreover, "Donald Trump RAC" this searching investigation case is also suspected of people-oriented, as we all know, a stem is "RAC-" is the result.

Cutts said before the media in a publishing this video does not contact Google to verify in advance, and ask: why don't you do some research before making a final conclusion?

SourceFed video here: click here to watch

从技术角度讨论：谷歌搜索有没有偏袒希拉里？ - 谷歌搜索,希拉里 - IT资讯

日前，一家名叫SourceFed的独立媒体发布视频指出，谷歌涉嫌操纵搜索结果屏蔽对希拉里·克林顿搜索负面关键词，以此在大选中支持希拉里。此消息一出，网友大呼“现实版《纸牌屋》”。

对此，谷歌方面否认了此事。

SourceFed是如何质疑的？

在这个被广泛传播的视频中，SourceFed称，当想查询希拉里的黑历史“hillary clinton crimes（希拉里犯罪）”时，键入前部分“hillary clinton cri”后，搜索框显示出的自动联想词是：

可见，谷歌的第二个和第三个选项指向了希拉里1994年提出的一个刑事改革法案，属于正向的搜索结果。

然而，搜索结果主要基于网友对于关键词的搜索次数，次数越多越可能成为自动联想关键词。比如，在必应搜索、雅虎搜索中，负面的联想词则要多些：

事实上，希拉里犯罪的事件本身更受关注，SourceFed查询谷歌的统计数据显示，相对于谷歌建议的关键词，“希拉里犯罪”的搜索流量更高：

同时，关于希拉里被起诉的露事件也是广大用户关心的，但谷歌联想词并未推荐：

也作为对比，必应和雅虎的结果是这样的：

除此之外，谷歌搜索关于其他候选人的负面消息的搜索结果却是和其他引擎比较一致的，比如对床破的搜索：

（民众对床破种族主义的质疑）

由此，SourceFed称，谷歌涉嫌操纵搜索结果，在特意保护希拉里的公众形象。

而如果行为属实，搜索引擎的这种做法有违道德，侵害了民众的知情权。

谷歌否认：修改了关键词提示，但未操纵搜索结果

面对此质疑，谷歌称：

“谷歌的自动联想功能并未偏向任何候选人，这些不实的质疑完全是因为对该功能工作模式的无知。自动联想词的生成基于多种因素，包括关键词热门程度等等。我们的系统会定期优化，用户的活动也会发生变化，所以自动联想词会相应地发生变化。此外，当与人名联系在一起时，我们的搜索引擎不会自动建议冒犯的或者毁谤的问题。

谷歌这样的回应也招致了部分用户的更多质疑，网友评论称：

这就是说，我就是弄了你咬我啊？

调查：指责有以偏概全之嫌

昨天，谷歌公司资深的高级工程师Matt Cutts在推特上回应称，这种指责是“低级错误”。Cutts指出，

SourceFed质疑的理由是使用谷歌搜索“hillary clinton in”自动联想没有出现“hillary clinton indictment”，输入“hillary clinton cri”也没有出现希拉里犯罪的搜索建议。但值得注意的是，事实上大部分用户在搜索时键入的都是“hillary…”，而非全名“hillary clinton…”。当搜索“hillary in”时，就会出现“hillary indictment”的自动联想。

小编实测：

与此同时，对于谷歌一方面也丑化床破的说法，也是非必然结果，上述举例中，输入的是“donald trump rac”，假若如前文一样，只键入“donald trump ra”时，按照搜索结果显示，谷歌也支持床破：

因为，用户对“特朗普离婚期间强奸（rape）妻子”一事的关注也比“集会（rally）”要高。

而且，“donald trump rac”这个搜索调查案例还涉嫌人为导向，因为我们都知道，词根是“rac-”的也就这个结果。

Cutts表示，媒体在发布这个视频之前，并未事先联系谷歌求证，并反问：为何不多做些调查再下定论？

SourceFed的视频在这里：点此观看