Information Overload: The best filter is relevant or popular?

IDG (Kathy) Beijing January 27, according to foreign media reports, the rise of social media has led to the proliferation of web content, but also raised the specter of widespread demand for information filtering tools. Popularity and relevance of the content is the most common filters, but they are the best?

Well-known tech blog ReadWriteWeb (ReadWriteWeb) Marshall Kirkpatrick (Marshall Kirkpatrick) to the three questions raised by the industry: In the future, relevance, popularity and filter what is the relationship? This three-class social networking software has been developed over the respondents were very different answer to this question. The following is a summary of Kirkpatrick article:

Popularity VS. Relevance

Before the advent of newspapers: the content of religious leaders of their choice to keep the content is not the point. But at the same time, they must pay attention to audience reaction. As Sy Safransky said, if Jesus was a pain in the ass, he would not attract audiences.

Newspaper Times: Editor's responsibilities and the needs of newspapers sold in the contradiction between, that is relevant to the content and the contradiction between popular.However, income earned on the popular content-related content is a subsidy.

The early Internet age: people can easily publish things, which makes the network appear on so many niche content, so that observers are worried that people are isolated in the information to justify. People seem less concerned about the popularity of the content?

Early social networking: Facebook and Twitter by the possibility of large-scale dissemination of content, so that publishers are starting to focus on link bait and other popular content.

Future? : The Internet is almost limitless amount of data need to select some of the content from the mass available to network users. So how relevant and popular according to the users to decide what content?

1 the meaning of relevance and popularity will change

Nick Halls Ted (Nick Halstead) is Tweetmeme (Twitter, popular content tracking service) and data mining company DataSift founder. He believes that the future for information filtering, relevance and popularity is very important, but the meaning of these concepts and the future is now different.

He believes that "changing the meaning of correlation. We used to think that relevance is the result of a query term and a match up, primarily through search engine keyword search to achieve. The popularity is based on the global behavior based single index (such as PageRank).

"But that will change both the meaning of development, social graph is changing the meaning. Now, correlation refers to a person's likes and dislikes - Facebook and Twitter have a lot on your preferences and patterns of behavior data. And now content is in accordance with the semantics of understanding and classification. So the relevance of the future will be what the contents of your preferences and match to decide.

"Popularity is changing the meaning, such as Klout and PeerIndex such services have proven social authority is becoming increasingly important. The first found that the 'top' of the company will become a great future, because this ranking means that the 'ad positioning ', which means revenue.

"Therefore, the future of the filter is still a combination of popularity and relevance. But it will be a new social indicators, in essence, you are not a 'search', but 'concern' to those with your match, no matter need to tell the computer.

"I think Google knows they need to quickly adapt to this situation."

2 different types of users require different types of filters

Feedly magazine reader interface released this week, an iPhone application, the industry that it is on the market the best iPhone feed reader. Feedly's co-founder and CEO Edwin Khodabakchian is one of the respondents.

Khodabakchian that network there are different types of users, they need different types of filters. Feedly content source in the process found the use of "relevance", in the process of filtering and prioritization in the use of the contents of the "popular." He said:

"We were divided into 3 types of users.

· Facebook users: they spend hours each day on the Facebook. This user likes and friends, the content is only for the interaction between them, pretending to be cool, feel themselves as a collective excuse.

* Passion-based user: They care about the theme and content of such users and the brand has a special connection. They like predictability.

· Twitter users and the blog of: their passion for information, desire for real-time.

"I think, for different categories of users, the filtering of information in different ways, for Facebook users, relevance and popularity is almost half and half, they will think: I can find some really interesting, unusual things to share with my friends and I look cool? for passion-type users, they have found the site and the content they trust the source, which makes things predictable. to provide them with the following three filters mixed: their favorite content sources, popular content, and recommended.

"Twitter is more interesting, because it has an implicit interest in maps, and it is based on the time. To know what is happening now, this very well ... ... from the sources you already know the content, which is very good .

"Our focus is food for thought you want to get the passion-type users. They are looking for a predictable experience, and discovered a little bit ......

"In Feedly desktop page where there is a 'you may also like the site' list. ... ... This is similar to Amazon's collaborative filtering on the website 'You might also like' list.

"In undertaking this collaborative filtering, we want a user and a source (or an author) to create links between, and not one to display the contents. Feedly about 70% of new users do not understand RSS or Google Reader. Personalized hardest part is that you only have 30 seconds to help them establish some relevant things, and you do not make their own, because their mind a blank. This process can not make people feel like is working.

"Even if the user Feedly very interested in a topic, but you just let them do such a little thing, they immediately walk away and give up, it makes me feel very surprised. I do not know whether the content source based on the value proposition , or related to the problem of information overload in the emotional colors. "

3, the most important user experience

Ouriel Ohayon is a "serial entrepreneur", his latest from a company recommended by the mobile application service Appsfire. Ohayon that the user experience is very important that the issue in the filter information, the user experience for both relevance and popularity does not seem so important.

Ohayon said: "The problem is not the algorithm or index, but rather user experience, you can develop the best filter, but it does not matter if test their tools and display their user interface is very bad words. you compare the Google Reader and Flipboard know.

"People are concerned about the appearance of new sources of information than off the back of the actual parameters. Of course, other factors or personal needs: social graph, the location (very important), freshness, interest in degree, but in general, users do not interested to hear this, they want a good experience, especially on the phone. "

As a technical person, I feel a little depressed. I do not agree with Ohayon's argument. I like to answer some of the more than 3 parts, but my own view is closest to the first two answer somewhere in between: The relevance of testing to determine the quality of the content source, a modest individual, and in the content level of popularity and content sources through different filter weights

I also want to pull the filter, "lid" playing himself playing its algorithm. But most people do not want to do so.

I hope in the future, we have the problem of information overload, dealing with a variety of ways, but for most people, and ultimately achieve this situation: What do people do the best content can be automatically sent to them, regardless of the "best "What is the way of decision.