Search Patterns with Peter Morville

New book coming out on search patterns from O'Reilly. Not due until next year.

Patterns - references Christopher Alexander "The timeless way of building" "Design Patterns"
Designing Interfaces by Jennifer Tidwell.
Yahoo! Design Pattern Library.
Search Pattern Library - flickr.com/photos/morville - examples of search.
Shows his pattern library with samples of best bets, faceted navigation, and so on. I should revisit this site, it appears to have gathered enough momentum and community to be helpful.

Behavior Patterns:
Narrow ->query->more keywords -> new results
Search & Browse & Ask all one thing to users.

Pearl Grow - find one good document and use data from that document to grow and explore other relevant and interesting documents. This is an expert search behavior.

Best bets - a few good starting points suggested by humans. Both broken out, and integrated into the search results. Can be used for query disambiguation.

Federated search - users don't know which database you hid the information within.

Faceted Navigation - multiple ways to search and browse through narrowing. Supports the way people think and behave within search. Serves as a map to the search results. Usually based on meta data. Usually dynamic. Usually showing the possible selections within the category and linking directly. Not the way we have done it. Need to use the search logs to continue to improve month over month. You need to include the number of results available within the value, important detail. "Scented widgets" very important. Need to see opportunities without being overwhelmed by them. Includes tagging within this area. Showed the VW UK site as an example of a graphic facets.

Auto-Suggest - because little things make a big difference.
Two types: Queries, based on own history or popular searches overall. Results - show the ones people have clicked on in the past within the results.

Structured Results - including structured information within the results, so apple stock results within a search on apple. Or a web site structure within the results list. We could take the resource search and use it to surface the sub navigation of the tool.

Speed is 100% most important thing. Google is successful because of speed of the
search.

Social search - bleeding edge.
Books & Authors - what do I read next?
Google and Amazon use popularity to bring the most popular results to the top. We need to do the same. Things people use, should be at the top. So we need to know what people use.
IBM - Enterprise 2.0 search within W3C web site. Talks to a their search results. They used tabs as a way of surfacing blogs. Did not work. Surfaced some of the content onto the page, and usage skyrockets. Integrated this information into the search algorithms. Using the use of data to raise the results.

Media Search - showed a site called oSkope. Microsoft image search on live - drag and drop, infinite scroll. Like.com find images by color, shape or pattern.

Mobile Search - as we get faster devices and better bandwidth, more becomes possible.

Spime search - Objects that know where they are, know their history. Example, query a rack of wine, as each is RFID tagged. Cisco Wireless Location Appliance is another example. Finding high value objects.

Google keeps expanding our understanding of what is searchable. everyzing uses automated speech to text to make a podcast searchable. And allow for search on it and leap to specific place in file. Audio and video become searchable.

Web 2.0 is the biggest Knowledge Management success in a long time.
Library example showing how search is connected to physical objects.

Apophenia - the seeing of patterns that do not exist.

Search is wicked problem. No definitive formulation, considerable uncertainty.
Complex interdependencies
incomplete contradictory and changing requirements
stakeholders have different world views and the problem is never truly solved.
It is a wider system, not just a particular interface or project

.

I highly recommend grabbing the podcast of these session and listening to it, again and again. There are many different ideas in this discussion that will be helpful in the future.

Presentation on slide share at http://slideshare.net/morville
His blog:
http://findability.org/
His email: morville@semanticstudios.com

Black Swan - we underestimate the ability of unexpected events to
completely change the world.

Building trust through restricted information: a case study of the hotwire.com air redesign

Melissa Matross Senior Manager, User Experience mmatross@hotwire.com
Trust is critical for any successful relationship. " A users willingness to spend time, data and money on a web site".
Brands she trusts - Nordstrom, Target, American Express, Apple.
Common elements: Service, Focused experienced, quality.
Hotwire is owned by Expedia, falls to one extreme - price sensitive customers who love travel and have no loyalty.
Business model: Opaque booking model. Suppliers have excess inventory and do not want to expose those prices to their loyal customers.
Air product conceals: Flight times, duration, number of stops, carrier. Savings of 30%. Core business model. Left with some level of distrust.
They have tried to overcome this distrust through:
Education
Testimonials
Direct comparisons
Putting work on the user does not help build trust.

When presented with a choice between opaque and retail more than 2/3 of the users chose opaque.

Dedicated customers were overwhelmed by options and did not find them useful. Tracked number of searches and purchase.

Users that indicated flexibility up-front were able to find lower priced alternatives and were more likely to purchase with hot wire.
1) Opaque = distrust.
2) shopping for travel = distrust & work.
3) competitor offering more data/options = more work & responsibility.
4) their users are more flexible than they know, if cost is lower.

Goals for redesign
Improve access to opaque inventory
Provide users with expected fare results while exposing them to lower fare alternatives
Present the trade-offs to the users
Improve user experience

Users were open to a series of trade-offs
Small - connections, schedule, carrier
Medium - changeability, alternate dates (leave on the 12 - 16), alternate airports
Large - products, destinations, flexible dates (any time in April)

Started to think about this as a travel agent. Simplified planning experience, knowledge, best recommendations.

Changed the interface to show the options and the impact of those changes, rather than just presenting the options. Gave them in terms of small, medium, large trade-offs. Gave more information about the opaque option, while still concealing what needed to be concealed. Gave and easy back option as well.

Layer only appears if there is a different alternative with a lower price.
Use this layer as a seducible moment - introduce the customer to lower price alternatives. Called it the speed bump model, wanted the user to pause and consider.

Limited choice - only 3 -5 best options. Give them 4 options - what they asked for, and 3 options based on trade - offs above.
Small trade-offs are always presented - opaque option and non-opaque options.
Anchoring & Framing - anchor on the exact choice, framed by the options.
User is able to understand the entire pricing context.
Did two usability studies: one as a low fi prototype, one as QA environment. Very positive reaction to the layer, very positive reaction to the search result.

Result
10% - 20% increase in alternate searches
50% - 55% higher opaque purchase rate for engaged customers.
10% - 20% increase in conversion
Garnered industry attention and good PR

Did these changes improve trust? Yes.

What is next?
Continue the model.
Positive feedback and conversion improvement.
Extending this model to the hotel and car searches.
Extending the model to other changes.

What can you do?
Understand what users really want, and what they say they want.
Need to understand what they say and what they do.
Understand how they make decisions and present trade-offs appropriately.
Identify what makes your offering unique.
Take risks.
Have fun.

How to be a UX One

Leah Buley from Adaptive Path presented probably the most "rock star" presentation. It might have even been right along the lines of Spools Keynote. The premise of her presentation is how to do UX only by yourself and she focused on the design portion of this process. On to the bullet points...
In her previous role she was the only UX person and had to do everything
She fell into a mode of moving blame around if something didn't work such as the development or functional requirements process
Went to adpative path and was handed a stack of postit notes and a marker and asked to sketch ideas and it freaked her out
Worked with other to generate design
She generates many ideas and then refines and removes designs
She now has 3 ideas for a team of one
Brainstorm a lot
Team (ad hoc) meetings
Pick your favorites designs
She took Evite and used their current page as an example of how she could work with it
Brainstormed and used conceptual frames works (she uses spectrums, 2x2 spectrums, grides and word associations)
She also keeps an inspiration library to review

Assemble an ad hoc team
Make sketchboards
Run template based workshops (we need to do this)
Decorate your space(people walk by and start talking)
Walk around and ask for opinion

Pick the best idea
Keep a point to stay in contact with and focus on that and meet that goal
Business needs are good
User needs are better
Business needs and user needs together = design principles
Design princples are not just one liners but a mixture that gets a strong selling point across.

Placemaking and IA

Definition– the process of creating plazas, parks, streets & waterfronts that will attract people because they are pleasurable
Not just functional, but pleasurable

We’re good at putting up buildings but bad at making places
Last half of 20th century was horrible for cities and communities – engineers were the cause of the problems. They focused on the objects rather then the good spaces for people.
Placemaking is
Community driven, visionary, function before form, adaptable, inclusive, flexible, culturally aware
Placemaking isn’t
reactive, design-driven, one-size-fits-all, etc
Some relevant principles
The community is the expert, create a place and not a design, you can see a lot just by observing, start with Petunias(iterative design, get design work done and redo it, don’t spend years), triangulate(make an object that two people can see and talk about – builds community – on content allow people to talk about the content), form supports function(in placemaking they talk about seating – how do you construct the seating so that you can have people comfortable, good vision, etc), you are never finished
IA Place Diagram
Sociability – Marriage proposals on twitter, divorces, feel comfortable interacting with strangers
Uses and activities – not lurkers who are hidden (ex. students who look up and research discussion topics to look brilliant), you want visible lurkers
Access and linkages – search, browse and ask
Comfort and Images – no porn, can visitors be seen in different areas
Washington Square Project
Maps and questions in the field, put pieces together, discuss the map and writing comments on their map – created a dense map that shows opportunities and problems
How to create successful markets workshop
Cheese shop… reflection in the display window, people can see each other at eye level talking to each other, should we build in “conversation seeing” (triangulation example)…they created a frame as well
South street bridge
Tables of teams, each had a facilitator, also had someone how could draw and give form to everyone’s ideas, had examples of other bridges, sketched out ideas immediately, very effective
Central Market
Talked about use of mobile devices in their market, how can info be available at the market that goes beyond brochure-ware, example of the carrots, need to balance - HR,physical resource and online

Probe I
Don’t think of it as design
Probe II
The IA is the facilitator, get people to see the same thing even though they are saying different things and think they are in agreement
Probe III
Know the context of the community, the large and small pieces of it… make sure the community is integrated and helping out each other, business owner associations built around the community… the IAs need to stick with the clients and business throughout and after the project… these sessions in the community are tough and passionate but it gets the truth out in the community and describing everyone’s difference.

Dilemma
Design – Steelcase hallways take people out of the flow and put them in cubbys, but people want to be in the flow of traffic and want to meet people that way (placemaking)

Inspiration from the edge . . .

Passionate about interface design.
How has the iPhone changed the web/desktop design? One example are new radio buttons.

Where do you get your ideas for interface designs? Generally people have "default thinking". So if you are designing a booking tool for a airline, people look at other on-line booking tools. Look at wundrbar - completely different experience - better. You need to look beyond immediate industry rivals for innovative design ideas. Another set of default thinking is to only use the standard buttons etcetera in the toolbox. Think differently about the interface.

Really good presentations
Architecture, Film & Mechanical objects - Dan Saffer.
Cinematic Interactions and so on - look at the slide deck that will be on-line.

3 quick comments:

With new technologies, almost anything is possible. Think about Adobe AIR,
Surface, Wii, Silverlight, Cloud Computing, XUL
Natural behaviors are
superior to learned behaviors. So, why are we pulling down to move a document up
with scroll bars?
Except when the learned behavior is better (makes me feel
better, makes me more productive) than the natural behavior.

Context: where this became important.
Project Crazy Quilt
Integrate dozes of existing applications, each with wildly different UI's, support power users & newbies, small business and enterprise, world wide on broadband and dial up companies.
How do we stich together multiple applications? Standard web thinking is to use tabs. Issue was, each app had it's own navigation. Saw the sugar interface and started thinking about the problem differently. Thinking that maybe the nav did not have to be persistent. Started to think about a "hub and spoke" model. Applications should not be navigation focused, they should be more focused on the ability to do things, task focused. Uses Club Penguin to show how the model of how the game works could be used to enable multiple applications to work.
Takeaway #1 look beyond the surface.
Looking at Photoshop, why could these actually be floating windows?
Takeaway #2 Think outside the (UI) box.
Deep customizability - games have very deep ability to establish things in new places, with new tools. We don't typically allow this as designers - we need to start thinking in this way.
How do we accommodate multiple workspaces?
Tabbed navigation? Doesn't work in his environment. Windows controls? Again doesn't work in this environment. Started thinking about scrollable work areas from a flash developers site that would allow for fixed areas of work, but with ability to move from one to another. Started looking at the iPhone for inspiration as well.
Are there better ways to display search results?
Started to split the screen with preview of documents on right. showed a flash site with weird calendar. This breaks out into 5 nodes of information. 1 Categories
2 sub categories 3 color 4 height 5 preview.
Takeaway #1 again - look beyond the surface.
How do we reduce the complexity of our applications?
Design with less space.
Think in conversations.
Songza as an example.
Focuses on what is important in each step of the process. So when you search for a song, you see the list. Once you select a radial menu opens with each thing you might want to do with a song: play, share, etcetera.
Getting away from persistent nav. Getting more towards how to enable tasks at the right point in the process.
Start looking outside the box more often.

ZUI - zooming user interface

The little UX that could...

This is a simple fairy tale that talks to the experience of being a UX person. You are not going to be the King, you are not going to be the Princess, so you have to find room for UX.

Given a simple development process - Definition, Back end Development, Implementation - the Princess (sponsor) will be involved in the definition, and be partly involved in the other two steps. The Wizard - back end developer - will be involved in the back end development based on definition and throwing it to implementation. The Witches - implementation team - will only be involved in implementing. As UX you will be involved in all steps in the process. So we have to create success without ownership.


Frequently someone else owns the product definition, and you will not be the only one who sees issues with the definition, but you will likely be the first person to say that we need to get the user involved.

Frequently, someone else owns the requirements. In the fable the little UX had to work with an incomplete set of requirements and expand them out to include the user. Usually, the project sponsor will push forward with development, even though you are not done. Be ready to go back and rebuild consensus.


Someone else owns implementation. They are going to be focused on implementation, not on the end user. They will be measured on their ability to implement the system, not how well the system works. You need to be the one to bring the purpose of the system back into the process.

Acceptance is not buy in. People will all love the user, until something causes them pain. You will have to go back, again and again to align
consensus and to bring the user back into the process.


Managing UX in the real world requires techniques and tactics that work up and down the conceptual ladder. Don't wait for a mandate. Concentrate on the idea that "I can make things better".


Always offer solutions to every problem you raise. Don't think your solutions are the be all and end all, but use them as a starting point for discussion. If you just point out the problem you will generate resistance.


Be ready to work harder than anyone else involved in the product.

Enjoy the successes, but don't take them at face value.

Don't assume that a successful project will have a positive effect on the next.

Folksonomy and Exploratory Search

Tingring Jiang & Dr Sherry Koshman
School of Information Sciences - University of Pittsburgh
3rd PhD student at University of Pittsburgh.
The material here is fairly simple overview of tagging systems. She compares and contrasts the various search behaviors with the activies in the social tagging systems.

Exploratory search and social tagging systems
Information architectures in exploratory search systems
Information seeking modes in social tagging systems

Exploratory search is the look for uncertainties in information need and information space. Looking with out a goal. More an attempt to understand the informaiton space. Typical activities are searching in a cognitive way, or browsing in a perceptual way. So searching is "teleportation" and browsing is "berry picking" in my terms.

Investigation into information seeking within social tagging systesm is typically exploratory. People are not looking up specific peices of information, they are instead trying to understand the system and how it will grow - trends.
She is taking a classic information sciences approach to folksonomy, finding it to be lacking in heirarchy and overly anarchic.

Four structures
Hierarchical classification
Faceted categorization
Dynamic clusters
Folksonomy

Hierarchical classification characterized by fixed non overlapping systems. Classic taxonomies.

Faceted Categorization characterized by a set of small hierachries that represent conceptual dimensions. Less human investment. Favors recognition over recall, making them easier for the user. This is limited to a fixed collection, unable to search dynamic material. These are similar to our filters in Beta II.

Dynamic clusters are generated based on the material retrieved. They are based on the clustering algorithms. Very automated. This is AQG in Autonomy and is in Beta III.

Folksonomy is characterized as being flat and inclusive. Based on anyone at any time in any language adding terms to the folksonomy. The benefits are inexpensiveness and responsiveness.

She gives a nice overview of the structures in a table. Worth including in our system.

She sees 4 types of behavior in social tagging systems
Browsing
Searching
Being aware
Monitoring

Still have the same 3 main elements: resources, users, tags. She then takes each in turn and shows how they interrelate and the behaviors of each.

This is a overview of tagging systems in general, not focused on using tagging in search or document retrieval.