Benign cycle of product interaction

The interaction between the user and the product can be seen as a two-way linear process, the input letter "D" on the phone and the interface output display is a short interaction, the completion of a shopping payment is long interactive, this is the use of a product process. The user experience is about the whole process of using the product, including the performance of the excellent user experience from installation to the time axis of abandonment, and the benign cycle of interaction.

Simply take the recommendation engine as an example to illustrate how each element works on the timeline:

Class。 Based on the group characteristics and the relevance of users, similar user preferences of other items are recommended to the user. Personal information (gender, age, income, etc.) that the user fills out can be used as an initialization recommendation when the user has not yet acted on the product.

Content recommendation. A single user selects an item, and the system recommends a similar product based on the metadata of the item. This is the user's individual behavior of the data filtering, when the user many times the behavior, the system can probably estimate the user's preferences. The user's historical behavior will continuously affect the follow-up recommendation, forming the interactive cycle between the user and the system.

Collaborative filtering. Discover the relevance of items based on user behavior. Content recommendation is a single user of the system data filtering, and collaborative filtering is based on multiple user behavior intersection results, so rely on other user behavior data volume accuracy.

The word association of input Method can also see a recommendation engine. When the letter "Da", the system from the thesaurus selected "Da" directory of the Chinese characters "da/da/play ...", the "big" word on the screen, the system Lenovo and "big" group word probability of larger words "home/school/...". These associative words may be selected from the dictionary as early as possible, if the user chooses the word "almost", "summary" will be weighted by the system, the next time to "big" Lenovo more forward display. User input phrase "Dagai", select "probably" on the screen, but also on the "almost" word weighting.

If the user input "Daniu", chose "Daniel" on the screen, and the system Word Library does not have this phrase, belong to the user to make the word. If the user input "Da" and "Niu" and on the screen, the System intelligence analysis "big" and "cow" have the probability of group words, it is possible that the next time the user input "big" word, the system will associate "cow".

Lenovo's interaction is a single user of the system Word library content filtering, the initial thesaurus can gradually form a personal thesaurus. If it is a cloud input method, the system can be real-time access to all users of the input words, the Word library update frequency faster, to meet the user's personality thesaurus can also update the latest popular words, this is a number of users of the coordination of the Word Library filter.

User-created words will reduce the probability of some words, and the system will even delete these words from the thesaurus. From the example of input method, we can see how the user behavior affects the system and other users, the historical behavior affects the new behavior and realizes the continuous interactive cycle.

Pure silver "Explicit content determinism" can be understood as: High-quality users to produce high-quality content, high-quality content to attract quality users, the content of the product to determine the charm and the field. From the point of view of system design, user behavior can be regarded as a part of system data, and excellent users ' behavior and quality content are homogeneous data. The frequent interaction between them is to optimize the organization form of the content data, and then to expand the generation and absorption of homogeneous data, the version iteration is a snowball data extension.

In this case, the product architecture needs to consider how to increase the amount of data, organization data and data expansion of the virtuous circle.

The Medal of the cut

As an example of mobile phone lbs, the method of increasing data quantity is divided into three kinds: User input, import data and merchant release information. Demand-driven users to actively use, the LBS Medal Incentive mechanism is to guide the user behavior, coupons are to stimulate user demand. The user's check-in behavior, if it is not available, can be considered as a constant input of spam information. However, the location of the check-in, the type of the check-in merchant and frequent procedures can be analyzed to facilitate the late referral of effective information to users and the organization of contact between users.

Each cycle will affect the subsequent product quality, "small run" the truth is also here, the user's feelings for the product is also increasing, but may also reach saturation. The use of watercress radio to choose to listen to music, the cumulative listen to nearly 20,000 songs, the radio guessed probability significantly reduced. The reason may be:

To understand the lack of organization of the product in the way of thinking guide schemata, Cycle is one of the primary concepts of product architecture, and the priority of product design is arranged with the timeline.

SOURCE Address: http://daichuanqing.com/i ... 5e7%258e%25af