On the information organization mode of UGC

Striking！ This article is rather boring, it is difficult to read patiently, early warning.

1,
About the organization of information, in the 1.0 era is nothing but artificial classification, the site set up the channel below the channel set column, the column below to set the Content List page. In the 2.0 era, the information explosion, it is clear that all can be manually collated. As you can see, there are mainly eight ways in which organizations work.

The first or manual collation, by the manual editing of the 1.0-style content home. Even 2.0 of the slogan has been resounded over the past five years, the average user is the most suitable for the traditional layout of the contents of the home page, looks neat and rich, clear classification, a good quality assurance. If you must rely on the algorithm to generate the home page, it is not not, but the trial and the cost of matching construction is too high. It's better to schedule one or two operators to update each day and achieve more operational intent.

The second type is sorted by Time dimension, often in a list page of massive information, divided into pure timeline (such as Twitter and feed streams), or back to the top ranking (for example, forums and Google Buzz). The former can guarantee fair exposure, which is applicable to places with higher value--the reply itself becomes part of content reading and effectively proves the quality and heat of the content.

The third is a variant of the time dimension, add new filter items outside of the time coordinates, such as clicking on the latest release of XXX, to maintain its quality, or interfering with the timeline with an interactive approach, such as the latest recommended content (the shrimp music channel).

The fourth is the interactive ranking, including the number of views/comments/recommendations/number of favorites ranking and so on. Not only the list of various time units (today/week/month/history ranking), but also divided into two categories of total ranking and increment. Among them, the reliability of the increment ranking is better, after adding the time exhaustion mechanism, it can avoid the unreasonable phenomenon of the big person.

The fifth kind is the user custom classification, is divided in the domestic acclimatized the tag, or the high-end user loved custom folder (for example watercress Bean row and Flickr exhibition hall).

The sixth is related recommendations, which may be based on user behavior or keyword matching, and may be recommended by the system, or by the user's own classification recommendations (such as Flickr photo sets and groups).

The seventh type is keyword search.

The eighth kind of situation is free and easy, does not organize the information, mainly relies on the follow to carry on the subscription between people.

2,
This log has been written for a very long time, it may be dragged for one months, a big change two times, write a waste of at least 2000 words, always dissatisfied. It was decided not to go around and show off, but rather to articulate the idea of how to handle the UGC information list.

This is a big problem, because the list of successful information is always accompanied by two elements: Category selection and quality control.

The forum uses the simplest and least effective product design approach to fix the problem. First, the senior management of the layout of the classification, to guide users to different sections to publish different types of posts, and then rely on the general Manager (moderator) to delete the irregular posts, rewards (top or add refined) excellent posts. So the content management pressure is completely decentralized to the operating side, the product designer does not hurt a paper brain.

Of course there is a background. A dense list of text titles, so that a small number of inferior main paste on the entire layout of the impact is small, and users are accustomed to quickly drag browsing replies, but also reduce the poor quality of the replies to the browsing interference. As long as diligent management, when the category and quality requirements of the content below a certain percentage, completely does not affect the normal operation of the entire forum. For pictures and video products, a screen mixed with one or two of inferior picture is miserable ...

A much more important advantage than visual pollution is that the forum is apt to inspire equality and positive discussion. Because of the active interaction, the topic is strong, forum may be the most easily formed public community products, thus safeguarding the sense of achievement and commitment of the duty manager, the high-end user driven up, relying on them to digest the huge operation and management costs.

For other types of products, due to the construction of the public community is congenitally deficient, can not find the good volunteers to work hard, in the information organization had to send hope in the use of labels, and rely on a variety of social recommendations to identify quality data. But besides tag is not popular at home, there is an awful paradox. When the viewer enters the latest information page, it's easy to run away with unfiltered bad content--a dump. Since no one wants to browse the latest information page, who will help you filter and filter the latest content?

Damn, it looks like a dead loop.

One of the solutions, first to the user screening, will be a reliable user to join the white list, there is no trial privileges. Other users of the content of the first trial and again, before the audit only in the latest information page display, do not enter the Select Information page.
Effect: poor. In my side, the insect pats the good user, may not be the portrait photograph is good, the scenery pats well the user, may not lomo also pats well. Although the use of a complex topic clustering white list algorithm, still does not solve the problem. Regarding the user's uncertainty factor too many, the white list user may not be assured, the White list user also brings the very big audit pressure.

Second, the use of some operations and algorithms to improve the quality of the latest information page to attract users to browse.
Effect: poor. Although the quality does improve, but still not enough as a user to enter the default page, if not the default page, the new content to attract users only about 10%, not up to the socialization recommended effective traffic requirements.

The third way to solve the problem is to learn the content recommendation function of Digg.
Effect: Very bad. Out of the cultural atmosphere of the geek community when Digg started, only the most boring and smug users would be happy to do it. Ask the Big blog operations team to know.

Solve the idea of four, let enthusiastic users to replace the administrator to audit content, many people voted to determine the results.
Effect: Only in the case of embarrassing sites have been successful applications, special background, it is difficult to copy.

The solution of the five, through the station dynamic and outbound transmission, to achieve a certain amount of browsing, on the basis of the socialization of recommended behavior to identify better content, to form their own select Information page.
Effect: It is almost the only effective way of UGC information organization at present. When a product has a strong personal center, your users are willing to get information through personal dynamics (or other subscriptions), or, like Youku potatoes, it's easy to spread information through the outside, and you'll get enough evaluations, referrals, and browsing numbers to help you make judgments. But here's the premise: often users focus on their personal space to access personal centers frequently. In other words, the personal center is the personal space first, the next is the relationship of the precipitation, must first do strong space, do strong relationship. Out of both, dynamic as the flower without roots. I've had a big loss here two years ago.

So, when you build a whole ecosystem of products from space to relationships, will it be possible to activate the social recommendation and generate a valuable list of selected information through the natural flow of information? Not so fast, but also depends on the form and algorithm of socialization recommendation. The simple approach is to browse the number of filters, in order to be more objective, it may also be recommended, share number screening, or scoring, evaluation number screening. No matter how the design, when your filtering algorithm is not open, users (especially in low-end users) will scold you play black box, the rules are opaque. But as long as a public algorithm, or someone guessed the algorithm, cheating immediately wildly popular.

I once met a group of users they each send a group of photographic work, no matter how mediocre, all the links to hundreds of people QQ group said, "Brothers, help a busy", and then tens of hundreds of people run to slam the top, even if the advanced users also swept among them. These QQ group almost became the public cheating group, each other 捧臭脚, when the operator questioned, the other side confidently replied: "So many people vote for me, why do you say I do not film well?" "No shame." Similar users in the number of photography enthusiasts, it is difficult to completely kill, only one of the Digg algorithm, reduce followers voting weight (Digg there are more ingenious algorithms can be used for reference)-but this rule becomes opaque, you will be scolded again ...

It's not a big problem to be scolded. Although the rules are open to the needs of community users, ignore them. Try again and again, always find the right filtering algorithm. The real trouble is that when a new user comes in, even a good user, he does not have any friends, can not get enough of the social recommendations, the work will not be able to access the selected information page, alone a poor. How to integrate new users into the information flow network? There are several countermeasures.

Method One, set up the new promotion bit in the public page, briefly introduces the newly added user.
Effect: Rotten. Who the fuck is nothing to do to some new head, not big wave beauty Maiden.

Method Two, the new works will be aggregated together to guide the new people to communicate with each other, holding group heating.
Effect: poor. Who's so sure the new guy likes to talk to new people? On the contrary, most couples are more willing to build relationships with senior users and integrate more quickly into the community. In addition this new work area must be mixed, browsing value is very small, stay time is very short, not worthy of attention.

Method Three, through some information matching means, both to recommend to the new people he may be interested in the old users, but also to the old users may be interested in recommending new people (the two forms are different). Help them build relationships quickly.
Effect: good. The premise is that this product has enough "user care" information coordinates (city/gender/interest/Subject preference/user level/similar background/neighboring IP, etc.), to achieve effective user matching and recommendation.

Method Four, open the group service within the product, equal to build numerous personalized small forum. Newcomers can join groups of interest to show themselves and make friends in the micro-public community.
Effect: poor. At home, a large number of low-end users of unbridled show, play and self-promotion, the strong defeated the group administrator's patience. Compared with the Success forum, the group is extremely scarce in the promotion means, it is difficult to accurately import the target users, most of them lack the clever operation management means. Or lonely boring to death, or low-end users to drive off the high-end users, leading to the decline of the entire group. Low-end users themselves can not produce quality content, it is difficult to form a content community, such as the African ant-like transit. Only the overall quality of the user group is high, the topic of strong products to establish effective groups, such as watercress and time, after all, is a few cases.

Method Five, the new user published the content of the operation Audit, on the one hand, filtering garbage, on the other hand, can also be a high-quality works directly into the selection of information pages, to help high-quality new users faster recognition.
Effect: good. The premise is that the cost of this operation, usually applicable to the publication of higher costs, quality judgments easier products. That's what I'm going to do. There are some algorithms at work that define "new user" and "junk user".

The introduction to the UGC Information List page is here.

3,
The Information list page is not the only way UGC information is organized. So far, list, dynamic, recommend, these three in different natures of 2.0 products have their status. For example, forum to list mainly, social networking sites with dynamic, I particularly want to talk about flickr--"recommendation" is its main information organization. From a good picture to a photographic world!

The story begins with every good picture. Where's the good picture? You click on the "Explore" this page, through the algorithm (browse number/favorite/author Grade/Publish Time Mix) extracted a series of high-quality photos, placed in disorder, beautiful and thrilling. Click in, because there is only one 640px large image on the page, you can easily notice the recommended position on the right side, including:
-All photos of the author
-Other photos of the author stored on the same album
-List of groups to which this photo is added

There are three key designs here. First, the main content of the page and recommended content in the right and left position, and the main content only one screen height, the left and right column width ratio of about 2:1, resulting in the recommended content is very easy to appear in the field of vision. To achieve this effect, Flickr is even willing to sacrifice the big picture experience, the main 640px browsing size to ensure the right column visual weight. It has benefited a lot from the loss.

The second is the large number of recommended content, because it may belong to more than one album, more than one group. On the right side of the photo, there are more than 10 recommended lists, each of which can be quickly expanded, scrolled around, and viewed directly from the author, album, and group in the form of a 75px thumbnail. The recommended effect is extremely strong.

Finally, the right recommendation is closely related to the main content on the left. I have this view that when the user's subjective control is stronger, his behavior becomes smoother and easier to sustain. Therefore, we should recommend the content of relevance as much as possible, and let the user understand the association, rather than place unrelated "essence recommendation." The latter may increase the probability of first click, but it is difficult to form a continuous click on the promotion bit. As an excellent case, Flickr author, album, Group Three dimensions constitute a reliable association, and provide the most love, the museum two types of secondary association.

In the endless extension of the chain, always follow the two logic: A, can make a good picture of the author must be able to make more good pictures. B, willing to include a good picture of the group will also include more good pictures. Therefore, the right side of the content promotion must be effective, while its position is eye-catching, interactive is friendly (Quick Preview + page refresh). As long as you find the first excellent picture, you can go along it, through the right column recommend a bit crazy, easy to find countless good pictures, countless good photographers, a whole photographic world door open.

This is how simple and beautiful, how amazing product design. Written here, only the "force" two words to aptly describe my mood. The weary author, however, plans to stop his pen and go to bed. Zhang Matchless schoolmate urges me to write this article, urges three times, spans one months. If he did not urge me, I would have to dig a hole shameful to slip away, at least late written.

Source: http://firecacada.blog.163.com/blog/static/7074376201072411591622/