The characteristics of information are multidimensional, which determines that information can be organized in different ways, access. Users generally have three modes of acquiring network information: one is the search mode; the second is the recommendation mode; and the third is the list mode. Each of the three modes has different characteristics, which determines their application in different user demand scenarios. We say that actually information is divided into two levels, the user side and the platform side. The so-called user side means that information is presented to users in a way that is convenient for them to recognize and access, and the so-called platform side means that information is stored in a form that is convenient for the platform to manage. Different information retrieval modes determine the different organization of retrieved information on the client side and platform side. For the client side, different user needs and information access scenarios determine different user demand points for information, so in order to facilitate rapid access to information, only the characteristics of the corresponding dimensions of the information need to be displayed. For the platform side, different information access modes correspond to different management and organization modes of the platform side (i.e., backend) for information. In other words, the main information retrieval mode of the platform will ultimately determine the visual presentation form of the information on the platform. The reason is that the three modes of information retrieval require different forms of information organization. The ultimate decisive factor is the user's scene requirements. In the actual use of the process because of the complexity of the information presented, a single information retrieval mode can not fully meet the presentation of information, and thus often the three with each other, but because of the user's access to information for the different expectations of the information presented in a way that determines the presentation of information is bound to a form of predominantly.
The key feature of this model is: fast and accurate. Users can pass a number of keywords, and quickly get the target information. Therefore, the mode is suitable for searching for users have a clear target information, favoring the precise target demand, the general user in the search for such demand content, pay attention to the accuracy of the content and timeliness, such as accommodation booking, travel taxi and so on. The most intuitive form of search mode is the search box, but it is not limited to this form, such as label filtering, keyword filter list and so on are different forms of search mode.
With the search mode corresponds to the need for related content labeling, that is, the use of the search mode of the content of the keyword extraction, information feature extraction, labeling categorization and so on. Particular attention should be paid to the search box search mode, because the user enters the site if the search box to search, that is, the user for the target content has certain expectations but does not exclude the user for the target information cognitive ambiguity, so appropriate keyword reminder is particularly important. As a matter of fact, the search box is an independent information portal of the platform, so the appropriate form of reminder can facilitate users to find the target information more quickly. The form in which the search results are presented depends on whether the user has a need for a choice of search results.
When the search is only as a site to facilitate the user to quickly search for the target information as a means (or information portal), the search form is only in the form of a search box, labels and so on, when the entire site information all need to have a search of the immediacy, accuracy, rapidity requirements, the search will be more than just a search box, labels as simple as the layout of the entire page needs to be With the information search results of immediacy, accuracy, rapidity of the layout, and the vast majority of information is all in the form of labeling in the background of the system for the organization, in this regard, the more typical representatives of the drip taxi, Goldmind maps, where to go.com, hungry, and so on. DDT Taxi, because of the user's high demand for immediacy and accuracy of information, the entire interface completely hides redundant information and presents only the unique results to the user. Other apps are not as extreme as taxi apps, but they also make it possible for users to quickly access search results in various ways. For example, the food ordering app Hungry Mans, which opens the app and automatically locates and presents information about nearby stores, is in fact a kind of automatic search for information based on geo-location tags.
Case 1: Dripping taxi
Compared to the timeliness (both sides), the need for vehicle selection can be basically ignored, which is the reason why the Dripping taxi page uses search with a simple model selection label and presents a single search result.
Case 2: Hungry.
The first consideration for dining is the dimension of geographic location, which is the default first search term. All the information presented is filtered under the dimension of geographic location, which is actually the search throughout the app. The presentation of the results is different from DDT, because users have the need to filter, so both the home page and the single content search results will set up a variety of search tags. Only the labels on the home page are set for all users, as well as the platform's requirements, while the user search results page shows the detailed filtering dimensions under the relevant content.
The key feature of this model is unpredictability. Users cannot know exactly what information they will eventually get. This model is suitable for recreational content presentation, because the user is browsing this type of information in the mind for the target information has no specific expectations, just trying to pass the time by browsing, and thus the recommendation model is more in line with the user at this time of lethargy, the expectation of the state of mind, such as today's headlines, NetEase cloud music and so on. However, given the current technical means, the recommendation model has many shortcomings.
With the recommendation mode corresponds to the need for a series of user behavior in the interface to collect and analyze, in order to portray the user's needs, summarized in line with the user's precise algorithm. However, because of the recommendation mode in the use of the scene under the user's behavior has a random nature, so the system needs to be found from the user's random behavior of the user's real needs, it is clear that this is not an easy thing, not to mention a single application platform, due to the user's content as well as the behavior of the singularity of the user's real needs to make the speculation of the user is also more difficult. Intelligent recommendation is the need for cross-platform collaboration, based on the user's behavior in different applications to summarize the user's real needs, it is clear that the so-called single-platform recommendation has yet to be perfected. The essence of the recommendation mode is still the search mode, only that the search mode requires users to retrieve information according to their own needs for information, while the recommendation mode is based on the background of the user's current operating behavior, automatically portraying the corresponding demand dimensions for the user, according to the portrayal of the corresponding demand dimensions and the content under the dimensions of the match. However, the problem lies in the lack of user behavior and the one-sidedness of the user dimension, which makes the so-called recommendation of the vast majority of platforms only stay in the simple association of related content according to user behavior (for example, I clicked a piece of information about dogs in the headline, and then a lot of dog-related information appeared in the process of browsing downward), and according to the platform's needs for the user to push the relevant high-heat content or content with commercial value. Currently, there is no application on the market that presents content in a separate recommendation mode, and the recommendation mode mostly exists as a form of content presentation, such as banner, related articles at the end of the article, and popular, the reason may lie in the so-called recommendation algorithms, the accuracy of which is doubtful, such as today's headlines, which boasts of personalized recommendations, although it boasts of recommendation algorithms, big data, and so on, but the content is mostly the same. Push content is mostly gossip, fun and so on, most users do not care about whether the push is accurate, because users are mostly in the pastime with the headlines, for the pastime of the user, accurate or not is in fact secondary, whether the user can get some kind of "immersive" lethargy in the process of pastime is important, the recommendation eliminates the user in different content lists. Recommendation eliminates the need for users to jump between different content lists, labels, entertainment gossip to meet the user's curiosity and do not need to think about it, with the almost subconscious operation of the decline of the update, perhaps this is the reason for the popularity of the headlines. Thus, the success of today's headlines does not stem from the accuracy of its recommendation model, but from its entertainment attributes make the vast majority of users for its accuracy requirements are not high, and its clever place is to match its content and interaction with the needs of the user, so that the user for today's headlines to form a mildly addictive (in the constant slide and gossip in the entertainment information, the user is easy to forget the time).
Given this uncertainty in the recommendation model, the vast majority of platforms use the recommendation model along with the other two methods. There is no such thing as an app that is entirely dominated by the recommendation model.
The key feature of this model is a full range of content categories. Instead of searching for information from the user's point of view, the list model is supposed to list all the information needed by all users of the platform from the platform's point of view. Therefore, from the user's point of view, it is not a good way to get information, but from the platform's point of view, it is an effective way to display a large amount of information. As a result, the platform uses a list mode along with a search mode and a recommendation mode to assist users in filtering information.
The opposite of the list mode is the need to summarize and organize all the user information for the platform, and through reasonable organization, make the information presented in a form that meets the needs of the majority of users, but also ensure that a single user can access the target information more easily. Although the list mode does not have the characteristics of the previous two search modes, the list mode is the most frequently used mode, the reason is that no matter which search mode will finally involve the presentation of information search results. Whether or not a list of information needs to be filtered with labels to help the user filter the information is a matter of judgment based on the user's need to select the information.
Like today's headlines there is no need in the "Beijing" this label and then add such as "life", "entertainment", "news" and "sports". " "Sports" and so on, of course, do not need because today's headlines for the whole country users, and then too much choice of labels in fact with the user with today's headlines for entertainment and recreation is contrary to the demand. Because too many labels to choose this time for the user to browse the lazy, curiosity psychology will cause an interruption.
Women's products like oxygen need to be displayed in the list of search results in the list of detailed filtering tags, because the user's filtering needs for this type of products is very refined.