GOMs keystroke-layer model for quantitative interaction behavior

Having been paying attention to the theory of quantitative experience, Sword Rainbow has benefited from sharing some of the methods of experience quantification in its previous blog post.

On the quantization method of the behavior layer I want to add the keystroke model in the GOMs method, which has the advantage of being more operable, and the designer can complete the measurement independently in a short time.

In the keystroke model, the user's interaction behavior is decomposed into several meta actions, each meta action is measured by a large number of tests to get an average time (see the table below), through the accumulation of these meta actions to get the interface design of the operation time required to verify and compare the pros and cons of various schemes. (depending on how time-consuming the individual action is, the typical value is sufficient for most contrast evaluations, such as the need for precise absolute time to refer to cpm-goms).

name and abbreviation typical value meaning keystroke (keying) K0.2 seconds keystrokes or clicks the mouse time to point (pointing) P1.1 seconds to point to a display device location time consuming (homing) H0.4 second hand switch between keyboard and mouse time consuming psychological preparation (Mentall Preparing) M1.35 seconds into the next psychological preparation time response (responding) R waits for the response time of the computer

The hardest part about using keystroke models is that you can't tell when users will stop to do unconscious mental activity. So 11545.html "> We need to refer to the following rules to define the timing of inserting m.

Rules for locating mental activities

Rule 0: Initial insertion of candidate M

Insert m before all K (keystroke), insert m before all p used for command selection, but do not insert m for p to select command arguments

Rule 1: Deleting predictable m

If the operator in front of M (k,p,h) can fully predict the operator after M, the M is deleted. For example, the purpose of your mobile mouse is to click http://www.aliyun.com/zixun/aggregation/26471.html "> Taobao home page, this time you need to delete by the rule 0 added M, then PMK becomes PK."

Rule 2: Delete m within the same cognitive unit

If a series of typing belongs to the same cognitive unit, delete all m except the first one, for example: input taobao according to rule 0 Insert m should be MKMKMKMKMKMK=6MK, because Taobao is a continuous input of a word so belong to the same cognitive unit, delete m should be mkkkkkk =m+6k

Rule 3: Remove the m before continuous non-terminal

If K is an extra delimiter after a cognitive unit, such as the separator of the command followed by the delimiter of the parameter, the previous m is deleted.

Rule 4: Deletion of M non-terminal as command

If K is a separator followed by a constant string (for example, a command name or an entity that uses the same every time), the previous m is deleted (the separator is habitually used as part of the string, thus eliminating the need for a separate m). But if K is a delimiter for a command parameter, or a string that may change, the previous m

Rule 5: Remove overlapping m

Do not count any m that overlaps with R (computer response time)

Simple keystroke model Calculation example:

Task Description: Search for T-shirts in Taobao, screen resources unrestricted, Taobao home has been loaded completed.

Meta Action decomposition:

Move your hands to the mouse: H

Point to search input box: HP

Click position Input: HPK

Move your hand back to the keyboard: Hpkh

Enter "T-shirt": hpkhkkkk

T-shirt in Sogou input method is the default, so just click on the space to confirm it. Add a bunch of hpkh if the Chinese input results are not in the default selected state.

Hit Enter to load search results: HPKHKKKKK

Page load into search results page: HPKHKKKKKR

Filtering in results: The best result is that the target product is ranked first in the result, and the user can directly point to the details page to complete the purchase hpkhkkkkkrpk. Assuming that the probability of this situation is P, then 1>p>1/7500 (75 results per page Total 100 pages) This probability is also by the sorting algorithm, the commodity belongs to the class objective standardization degree ... And so on complex factors, so the operation after the completion of the query will not be covered in this example.

So the final expression we arrive at is: HPKHKKKKK

Add initial m according to rule 0

Expression is: HPMKHMKMKMKMKMK

Remove excess M based on rule 1-3 expression: hpkhmkkkkk=2h+p+6k+m+=0.8+2.2+1.2+1.35+2=4.55

If you use the automatic complement function of the search input

(Input method defaults to e-text) expression can be adjusted to: hpkhmkhpkpk=3h+3p+4k+m=6.65

By comparing the results of two expressions, we can conclude that the input efficiency of automatic complement is lower than that of full text input under some conditions.

The advantage of the quantitative approach is that it can turn endless arguments into calculations and come up with compelling conclusions that are generally accepted.

The GOMs keystroke model has great limitations, it is more emphasis on the physical level of measurement, and Taobao users in the process of shopping experience screening; contrast; impression; experience; the complexity of the psychological factors in the interactive behavior level has not been relatively accurate measurement methods to be further study.

Bibliography: "The Humane Interface"

Source: http://ued.taobao.com/blog/2010/05/27/goms/