tomires/nur.md

## nur.md

      
    Raw
  

              nur.md
            
          
    1


human-computer interaction - the process of design, implementation and evaluation of interactive systems from a PoV of the user

interaction meaning two-way communication


user centered design (UCD)

design over user needs
involvement of users in the design process

preparation stage

user research - process of gathering tidbits of data about the users, domain, etc.
user modeling - creation of context


design phase

sketching - communication of the main idea to our users
rapid prototyping - creation of prototypes of varying complexity


implementation
evaluation
sales


the reality is far less rigid, stages often blend into one another
as part of SW frameworks

waterfall - design stage
mini-waterfall - implementation stage


users want

to be in control
to focus on the task at hand
not to feel stupid
not to think too much


attempt to cater to different kinds of users, but do not attempt to fill every niche
usability

recall - (long-term) memorability of the environment
efficiency - the concept of flow
simplicity - controls are easy to learn
low error rate - prevent the users from making them in the first place, give valuable feedback
user satisfaction


terminology

task - set sequence of activities
action - one step performed as part of a task
story - short narrative description of user activities written by user
scenario - informal narrative story, personal, natural, not generalizable
use case - set of scenarios
storyboard - a series of sketched key frames


2


hierarchical task analysis

visualizes division of tasks into subtasks

gain insight on structure of tasks (sequence of atomic steps) needed to achieve goal
can be used to capture user requirements, in UI design (sequence of queries, menu design, etc.)


tree structure (goal - tasks - subtasks - steps) / textual summary
plans - optimal tree traversal to achieve sth (mention only leaves)
level of decomposition depends on individual task and importance of doing so


sketch

crude, hand-drawn
describes ideas, arouses creativity (for coming up with solutions)
do not confuse with low-fidelity prototype, sketches should not include application screens, just the general flow of the experience

suggest x describe, question x answer, explore x refine, propose x test


mockup

a crude and incomplete version of prototype
includes only important states, elements
too basic for user testing


design studio

workgroup for brainstorming ideas
should include all stakeholders
end result - sketches


3


lo-fi prototype

short time frame (hours, days)
enables for quick edits (fast workflow)
does not use target devices
prerequisites - user research, task analysis (HTA, scenarios, mockups)


4


information architecture

domain entities and their relations
HTA, card sorting


hi-fi prototype

longer time frame (weeks/months)
should feel like the final product, essentially a facade
should adhere to HIG of target platform (UI design, input methods)
does not have to include app logic (usage of mock data, WoZ technique)
needs to contain all key parts of the application


evaluation process

summative

at the conclusion of SW development process


formative

throughout development (continuous evaluation)


5


communication between controller and view (MVC) - complete separation is often violated in practice
dialog

typically not structured
modeling - graph with conditional transitions
typical features

sequence of actions
fixed parts
variable parts


dialogue models

state transition diagrams (STN)

pretty much a FSM
transitions can be conditional
can be hierarchical (for sub-menus, etc.)
ATN - uses registers
we can use graph theory to find shortest paths and check validity of the final graph

all states should be reachable (strongly connected graph) easily (subjective)

a tour across all nodes can help determine complexity of the solution


we should be able to get back to previous states
automatic tests and even semi-automatic generation!


concur-task trees (CTT)

introduces symbols

user task (think something through)
interaction task (select something)
cooperation task
application task (computer crunches through shit)
abstraction task


introduces operators

binary

>> - enabling
[> - disabling
|> - interruption
[] - choice
||| - concurrency


unary

T* - iteration
[T] - optionality


petri nets (PN)

utilizes token shift
used programming constructs

concurrency
synchronization
sequencing
selection
control branching
looping


monolithic architecture - presentation, dialogue, application (PDA)

as with MVC, a change in one part propagates to other parts


feedback

lexical - movement of mouse
syntactic - menu highlights
semantic - sum of numbers changes

often slow, the user demands fast reactions
use syntactic feedback to alleviate this issue


UI models

seeheim

presentation

generates output
receives input events (lexical parsing)


dialog control

parsing of tokens into syntax
maintains state


application interface model (AIM)

interface between UI and the rest of the software
semantic feedback - checks validity of inputs


6


perception

bottom-up - focus on pattern building
top-down - dependent on attention, focus on a single part
low-level features analyzed

form (size, rotation)
color
motion


pop-out effect

object is sufficiently different from its neighbors
should be used for important objects
use primary features (size, rotation, color, motion)
do not combine multiple features at the same time
objects should retain their look (e.g. button should not look like an advertisement)
stylistic consistency leads to comparatively longer visual search process


biased feature competition

we already have a prior knowledge of how the object looks like


gestalt principles

proximity
similarity
closure
symmetry
continuity


use color coding, familiar shapes

problem with interference of specific cognitive processes

ex. with colored text


use maximum of 4 colors
beware of cultural context
be consistent (for the most part)
use fall-back (e.g. for colorblind people)


icon design

findability (placement)
recognition (floppy disc -> save dialogue)
information scent
attractiveness


8


cognitive modeling

deals with producing computational models of how humans perform tasks and solve problems
used to estimate task and training time and answer design questions
model human processor (MHP)

short-term memory processes perception and motor movement
cognition is processed in both short- and long-term memories
processing can be serial, parallel
parameters

processor - cycle time
memories - storage capacity, information decay, information type


memories

sensory - "buffer" for stimuli - visual, audio, tactile
short-term - 70ms access time, 7s decay, stores 2-4 chunks
long-term - 100ms access time, slow decay, large capacity

declarative information - facts, data, events
procedural - how to do stuff (if - then)


stimuli goes from sensory memory to short-term memory via attention
storage in LTM

via rehearsal
the more time we spend rehearsing, the more we remember
spreading learning over different time periods helps
presence of structure, knowledge of meaning (domain) also helps


forgetting information in LTM

interference with new information (German v English) or replacement
gradual decay
memory is selective and can be subconsciously affected by mood


information retrieval from LTM

recognition - helped by stimuli
recall - no outside help, spontaneous


cognitive processes

selective attention

focus on one object
failure in object formation
switching attention

takes 100-200ms
Cocktail party effect


learning

parts

memorization
understanding concepts and rules
acquiring and automating motor skills


can be helped

structure
similarities with something else
repetition


is hindered by

conflicting previous knowledge
inconsistency


problem solving

deductive (rule-based, if - then)
inductive (generalizing from previous cases)
abductive (causation)


language

rules, patterns


models of human behavior

GOMS

predicts user performance
can be used in design stage
parts

goals

what we want to achieve
can be decomposed into subgoals


operators

actions (as atomic as possible)
ex. move mouse to icon, press button
we assign a time value to each operator


methods

sequence of operators to accomplish goal
ex. move mouse, depress button, drag mouse, release button


selection rules

decision tree
if there are multiple ways to achieve goal, decides which method to use


output

total execution time
rank ordering
procedures executed


pitfalls

we presume the user is trained
we can assign a time value to every operator
we presume no outside interference


keystroke-level problem (KLM)

based on GOMS, inherits basic characteristics
used operators

K - keystroke
P - point (mouse dragging)
H - homing (movement of hand from default position)
D - drawing
M - mental activity
R - response (output / system)


time duration is provided for user operators
experienced users may be capable of multitasking (mental operations run in parallel with other actions)


power law of practice

time to complete task on n-th try
Tn = T1 * n^(-a)

T1 - time for first trial
n - number of trials
a - learning rate (0.2 - 0.6)


applies only to skilled behavior (perceptual, motor)


hick-hyman law

time it takes for the user to make a mental decision
T = a + b * log2(n + 1)

a - basic processing time
b - additional time per choice
n - number of choices


fitt's law

measures movement time for selection tasks

the time increases with distance, decreases with size of target


T = a + b * log2(2d / w)

a, b - constants

need to be acquired by testing and subsequent statistical analysis


d - distance
w - width of shape (in direction of motion)


9


guidelines set rules, these can be simplistic, abstract

HIG are meant to enforce a unified design philosophy across elements of a specific ecosystem
ISO guidelines - software ergonomics, haptic interactions, displays, etc.


patterns are context specific

tabs - selected tab should be clearly highlighted, they are used for alternate views within same context
sliders - used when precision is not required


11


types of impairment - hearing, visual, motor, cognitive
offer alternative means of input, output
heuristics

content parseable by screen reader (respect HIG)
alt text
use headings
immediate validation


focus on user's abilities rather than limitations, design from scratch

12


intelligent system capabilities

reasoning - derivation of new knowledge based on existing knowledge (transitivity in graphs)
learning - improvement of behavior
adaptability to changes in environments


intelligent user interface

link between user and intelligent interactive system
natural method of communication, accessible (UI adapts to user's capabilities), low cognitive load (UI personalization)


computer mediated interaction

synchronous x asynchronous
local x remote