Dr. Tony Russell Rose, FBCS CITPblog.it-faggruppen.dk/.../Designing-the-Search...1.pdf · How it works •User interface design • Information architecture •Interaction design

10/07/2013

1

Dr. Tony Russell‐Rose, FBCS CITPUXLabs London UKUXLabs, London UK

Tony Russell-Rose, PhDDirector UXLabsDirector, UXLabs

20 years of experience of applied research & technology innovation at Microsoft, Canon, Reuters, BT Labs, HP Labs, Endeca

Vice-chair of the BCS Information Retrieval specialist group. Chair of the IEHF HCI group. Founder of Search Solutions conference series

PhD (natural language interfaces), MSc in human-computer interaction, first degree in engineering (human factors)

3 patents on search user interface design. 60+ scientific and

Industry Expertise

Business Intelligence

Government

Media & Publishing

Healthcare

eCommerce

Domain Expertise3 patents on search user interface design. 60+ scientific and technical papers on information retrieval and user experience

Honorary Visiting Fellow at City University London (Centre for Interactive Systems Research)

Blog: http://isquared.wordpress.com

Domain Expertise

Speech & language interfaces

Information architecture

Info visualization

2UXLabs: User Experience Research + Design + Analytics ‐www.uxlabs.co.uk

10/07/2013

2

Time Topic

00 00 I d i d bj i00:00 Introductions and objectives

00:10 Understanding information-seeking behaviour

00:50 Formulating the query

01:30 BREAK

02:00 Displaying results02:00 Displaying results

02:45 Faceted search

03:40 Conclusions & wrap-up

04:00 CLOSE

3UXLabs ‐User Experience Research and Design ‐www.uxlabs.co.uk

The dimensions of search user experience

4

10/07/2013

3

Where should the Faceted Navigationmenu be located and ghow should it be organised?


How should we present mixed t t lt t h l di content results to help users discover

the most relevant and useful items?


10/07/2013

4

GoalUser Type Search ModeContextGoalUser Type Search Mode

Analyze Compare Evaluate Explore

Context

Spatio‐temporal Social Physical Information Assets

The latest Harry Potter book… Shoes to match my interview suit …

Electronics Engineer Purchasing Agent Novice Shopper Technical Enthusiast p o e

Locate …

o at o ssets … An affordable

entertainment system …

ec ca t us ast …


Users vary in their technical and domain expertise

EXPERTISENovice

Uncertain

Expert

Confident

Novice Shopper

Tech Enthusiast

News Analyst

Casual News

Reader


10/07/2013

5

User goals vary in scope & complexity

I need to find part number 2335456

SCOPEFocused Broad

I want to buy the new Harry Potter book

I want an affordable entertainment system for my family

I want to buy shoes that match my interview suit

I need to discover compatible parts that work for my assembly

I need to understand part obsolescence & manage our risks

Simple Complex

Electronics Engineer

Novice Shopper


COMPLEXITYSimple

Homogeneous

Highly Faceted

Diverse


10/07/2013

6

Marchionini 2006Marchionini, 2006

• Lookup

• Learn

• Investigate

Morville, 2010

• Quit

Ellis, 1993

• ChainingQuit

• Narrow

• Expand

• Pearl‐growing

• Pogo‐sticking

• Thrashing

• Browsing

• Differentiating

• Filtering

• Monitoring

• Extracting

• Verifying


Lookup

• Locate

• Verify

M i

Learn

• Compare

• Comprehend

E l

Investigate

• Explore

• Analyze

S h i• Monitor • Evaluate • Synthesize


10/07/2013

7

How it works

• User interface design

• Information architecture

• Interaction design

How it looks

• Visual design

User Research

g

Design Methodology

• User‐centred design

• Agile development


Identify need for user-centered design

System satisfies usability and business goals

Specify context of useWho will use it and under what conditions?

Specify requirementsWhat are the business and user

Evaluate designsTest the design with users

g

gSUCCESS!

14

* ISO 9241-210: Human centred design processes for interactive systems

What are the business and user goals?

Produce design solutionsPrototype the design

UXLabs ‐User Experience Research and Design ‐www.uxlabs.co.uk

10/07/2013

8

UXLabs: User Experience Research + Design + Analytics ‐www.uxlabs.co.uk 15

Principles and practices


10/07/2013

9

Think affordance:ff

form should follow function

Best practices:

Be consistent

Pre‐locate the cursor

R i h Retain the query

Use placeholder text & prompts


Facilitates known‐item search

Requires domain expertise

Risks over‐constraining

Best practices:

D f lt t ‘ ll’ Default to ‘all’

Apply a fallback

Echo the scope


10/07/2013

10

Can be applied as fi trefinement

In faceted nav menu Or as toggle to existing search box

Best practices: Display refinements in breadbobreadbox

Disable when not applicable

Remove current query


A flawed conceptp

An extension of parametric search

An effective search experience scales with the userthe user


10/07/2013

11

Autocomplete Accurate & efficient data entry

Controlled vocabulary: recognition not recall

Known‐item search AutosuggestAutosuggest Novel query reformulations

(relatively) unbounded list

Exploratory search


Equivalent to ‘best bets’q

But prior to query completion

Product promotion

Can be applied to queries


10/07/2013

12

How do you repair an unintended ‘detour’ yet y p ysupport serendipitous discovery?

Search engines cannot read minds

Or determine user intent

Anchoring bias leads to ‘thrashing’


Did you meany

Not ‘failed’ as such

Potential typos

Return a more productive result set


10/07/2013

13

Uses query reformulation q ypatterns

May not know true intent

But can rule out certain interpretations

Avoid zero results Avoid zero results


Over‐constraining is gcommon

esp. for zero SERPs

Identify productive permutations

Di l t hi Display non‐matching elements

Preserve diversity

Quorum‐level ranking


10/07/2013

14

Query clarificationy

Find similar

Associated concepts




10/07/2013

15

Dual purpose:p p

Communicate richness & diversity of result set

Convey detail of individual items

Information scent vs. screen space

‘Pogosticking’


A balance is required …q


10/07/2013

16

Query‐oriented isummaries

Show keywords in context

Highlight matching terms Use previews Support for verifying


Not just passive ‘containers’j p

Articulate the aggregate response to an info need

Google includes 12+ separate features

Input

Informational

C l Control

Personalization


10/07/2013

17

Basic principlesp p Display current query (in editable form)

Communicate navigational state

Display the number f ltof results

Support diversity▪ esp. for vague or ambiguous queries


Select a layout that t h th d t matches the data

and the task, e.g. List view Gallery view▪ Explore, locate

Parametric viewl▪ Analyze, compare

Map view▪ Geospatial or conceptual

Provide user control


10/07/2013

18

Goal: promote pexploration & discovery

support a multiplicity of formats

Lists tiles panels Lists, tiles, panels


See Also Featured results ‘Best bets’

Natural language obeys a power law

Provide known good Provide known good matches for popular queries


10/07/2013

19

Clearly indicate when there are no search

lresults

Provide meaningful navigation options top searches, featured results, most popular results

P i i b h Prevention is better than cure auto‐suggest query expansion partial match results


Sorting ~= filteringg g Not just relevance Category‐specific Drop‐down or tabular?


10/07/2013

20

Search engines gcannot read minds

Encourage category selection

‘Clarify, then refine’




10/07/2013

21

Facets are categoriesg Categories represent the properties of objects

Categories have values that describe the object

But not all categories can serve as navigable But not all categories can serve as navigable dimensions


Facets define values in multi‐dimensional space

Facets provide multiple paths to any given item, e g a recipe defined by:e.g. a recipe defined by:

Course

Cuisine

Cooking method

Stephanie Lumieux“Designing for Faceted Search”


10/07/2013

22

What are the typical facets of wine?yp



10/07/2013

23


Well‐chosen facets h ld bshould be:

Exhaustive

Consistent

Orthogonal

Should induce a high‐entropy distribution in the result set good bad


10/07/2013

24

Layouth Position on the page

Default state Open, closed, hybrid

Display formats Links, checkboxes, sliders, etc.

Scalability Showing more valuesShowing more values

Navigation & wayfinding Orientation in complex information spaces

Interaction models How facets behave


3 common configurations:3 g

1. Left (vertical)

2. Right (vertical)

3. Centre (horizontal)

Hybrid


10/07/2013

25



10/07/2013

26



10/07/2013

27


Three primary options:p y p

1. Closed by default

2. Open by default

3. Combination of open & closed

Context‐dependent


10/07/2013

28



10/07/2013

29



10/07/2013

30

What are the advantages and disadvantages g gof the following layout configurations?

Advantages Disadvantages

Layout: Left

Layout: Right

Layout: Centre


Layout: Centre

State: Open

State: Closed

State: Combination

Fundamental principle:p p Match the display format to the semantics of the data

Primary formats: Hyperlinks

Checkboxes

Sliders

Input boxes Also: colour pickers, tag clouds, data visualizations, etc.


10/07/2013

31



10/07/2013

32



10/07/2013

33



10/07/2013

34


What display formats would you use for the p y yfollowing data?

Type Valence Display Formats

Binary

Flat Single‐select

Flat Multi‐selectOR


Flat Multi‐select AND

Hierarchical Single‐select

Hierarchical Multi‐selectOR

Continuum

Interval

10/07/2013

35

Summary and wrap up

69

Adopt a principled approach to designp p p pp g

• Know your users, their goals, context and search modes

Look at search holistically

• Think beyond findability to analysis and sensemaking

Study other search applicationsy pp

• Different contexts require different solutions

Take advantage of online resources

• Apply proven design patterns and principles


10/07/2013

36

Search usability resources

71

Faceted Search Daniel Tunkelang, 2009

Search User Interfaces Marti Hearst, 2009

Search Patterns Morville & Callender 2010

Designing Search Greg Nudelman (2010)

Information Architecture for the WWW Rosenfeld & Morville (2006)


10/07/2013

37

The Design of Everyday ThingsN Norman 2002

Designing Interfaces Jennifer Tidwell (2005)

The Design of Sites Van Duyne, Landay & Hong (2002 / 2006)

Designing Web Usability Jakob Nielsen (1999)

Don’t Make Me Think Steve Krug (2005)


Yahoo! Jenifer Tidwell

Endeca Peter Morville


10/07/2013

38

Designing the g gSearch Experience: the Information Architecture of Discovery

Tony Russell Rose Tony Russell‐Rose and Tyler Tate

Morgan Kaufmann, Dec 2012


Tony Russell‐Rose, PhDVice chair BCS IRSGVice‐chair, BCS IRSGChair, IEHF HCI Group

Web: http://uxlabs.co.uk Email: [email protected] Blog: http://isquared.wordpress.comg p // q p LinkedIn: http://uk.linkedin.com/in/tonyrussellrose Twitter: @tonygrr


Dr. Tony Russell Rose, FBCS CITPblog.it-faggruppen.dk/.../Designing-the-Search...1.pdf · How it works •User interface design • Information architecture •Interaction design

Documents