网站制作内容的推送、寻找、分类、筛选的方式有以下四种形式分类，或多或少的，同时并存。 Website content push, search, classification, screening way has four kinds of classification, more or less, at the same time. 一、用户自选(Users customize) 这一种方式用户具有最大的主动权，网站通过设计清晰的分类导航和站内搜索系统，让用户通过自己的意愿，在网站上寻找到他需要的内容信息。 This way the user has the initiative the biggest, site classification through navigation and search system design is clear, let users through their wishes, to find he needs information on the website. 优点：条理清晰，简便有效。 缺点：是否能找到精确的信息，一方面取决于用户对信息的归类方式和搜索关键词的组织，一方面取决于网站对信息的归类和关键词的组织。如果两方对同一内容的需求和理解不一致，，一个信息内容，网站归如A类别，用户认为B类别可以找到，那么一定是无功而返。那么这个信息应该打A,B两个Tag，归入两个类别。搜索也是同样的缘故，要在搜索的索引中将可能被用户查询的信息都提供完全。最后，大量的分类和冗杂的搜索信息积累下去，反而又从精确变成了模糊。于是人们发现，各大搜索引擎上的关键词越来越长。 Advantages: clear, simple and effective. Weaknesses: if you can find the exact information, on the one hand depends on the user of the information classification and search keywords organization, on the one hand depends on the website for information classification and keyword organization. If the two party needs and understanding of the same content, inconsistent, an information content, such as web site to A category, B category of users that can be found, then it must be failed. This information should be playing A, B two Tag, fall into two categories. Search is the same reason, in the search index may be a user query information provided is complete. Finally, classification and miscellaneous large search information accumulation, but also from the exact into fuzzy. So people found, each big search engine keyword more long. 总结：作为最基础的方式，没有网站不会提供这个功能，至少都有对信息的分类导航。 Summary: as the most basic way, no site does not provide this function, at least the classification of information. 二、审核匹配(Audit matching) 审核是指对内容进行审核筛选，诸如通过评分体系，浏览量，点击率，时间和等等其他一些因素，给与大量同类信息进行排序和权重区分，来判断这个内容对用户的需要程度高低，在分类列表或者搜索结果里，将最可能的用户需求信息最先或者最明显陈列给用户。 Audit refers to the audit selection of content, such as by scoring system, the quantity that browse, click rate, time and so on some other factors, sort and weight to a large number of similar information to distinguish, to judge the content of the needs of the user level, in the category list or search results, user demand information will most likely the first or the most obvious display to the user. 优点：为第一种方式做补充，通过规则对信息进行排序和过滤，降低大多数用户的搜寻成本。 缺点：这是一种少数服从多数的匹配方式，即默认需要信息的用户是“大多数用户”，忽视和无视“少数用户”的需求。而评分体系也有体系的问题，让用户进行评分，并不会产生理想化分数，专业评分又缺乏性。关于商品分数的问题，我在另外一篇文章里做过讨论——现有商品的评分体系不完美另外，YouTube也得出了结论说，评分的分布并非理想化的。 Advantages: as a supplement for the first way, through the rules for sorting and filtering of information, reduce the most users search cost. Disadvantages: it is a way of matching the minority is subordinate to the majority, which is the default need information of the user is " most users, " and not " a small number of users " demand. The scoring system also has the problem of the system, allowing users to score, and produce ideal will not score, score and the lack of professional. About the commodity the score, I in another article discussed -- existing goods scoring system is not perfect in addition, YouTube also draws the conclusion that, the distribution is not ideal score. 总结：在这个匹配体系里，最大需求被最大满足，最小需求被无视，从损益综合来讲是最优解，但是当这个社会的个性化需求越来越强烈的时候，长尾上的用户需求就没有明显的顺序了，完全是一小撮用户有一小撮不同的需求，而且他们加起来就是一个不可忽视的长尾，怎么给他们推送内容信息呢。 Summary: in this matching system, the maximum demand is the biggest meet the minimum requirements, be ignored, from the income in general is the optimal solution, but when the personalized needs of more and more intense, the long tail of the user needs no obvious sequence, is entirely a handful of users have a handful of different needs, and they add up to is a can not ignore the long tail, how to give them a push content information. 三、个性化匹配(Personalized matching) 通过多维度的了解用户的偏好，对用户的信息需求进行匹配。喜欢喝果汁的人通过“苹果”这个词首先查看的是“苹果汁”相关信息，而一个“果粉”通过“苹果”这个词首先找到“苹果电脑”。怎样去定义一个用户个性维度，一般有两种方式。 Through the multi-d