Joe Lamantia.com

I’ve posted the slides from my UXLX talk on the Lan­guage of Dis­cov­ery. Thanks to a few days spent fea­tured on the slideshare home­page, they’ve clocked over 60,000 views in the past week!  In com­bi­na­tion with the buzz from the audi­ence for the talk, I think this shows there is broader aware­ness and appetite for answers to the ques­tion of how design­ers will make big data acces­si­ble and ‘engageable’.

From the prac­ti­cal per­spec­tive, if you’re look­ing for a way to describe dis­cov­ery and sense mak­ing needs and activ­i­ties, there’s no bet­ter resource than this.  And the LOD is well-grounded from the method­olog­i­cal and research per­spec­tives, hav­ing roots in HCIR, cog­ni­tive sci­ence, and a num­ber of other aca­d­e­mic dis­ci­plines that con­tribute to the toolkit for under­stand­ing human inter­ac­tion with infor­ma­tion and dis­cov­ery activity.

I hope the lan­guage of dis­cov­ery is part of that big­ger pic­ture of how cre­ators of inter­ac­tions and defin­ers of expe­ri­ences shape the new tools peo­ple use in the Age of Insight.

Also, the Lanyrd page for the talk aggre­gates the slides, sketch notes, and point­ers to some other resources.

Sketch­notes from my UXLX talk are posted. Thanks to the crew at Lives­ketch­ing for cre­at­ing these, and shar­ing them (this photo is cour­tesy of flickr user visu­alpunch).

As I’m sure you can see by the level of den­sity, I was mov­ing quickly to cover a lot of ground…!

 

The com­plete set of sketch notes from UXLX is avail­able as a set on flickr here.

I’ve just con­firmed that I’ll be pre­sent­ing a light­ning talk at this year’s UX Lis­bon, in May — I’m excited!  There’s a great lineup of UX speak­ers, and I’m look­ing for­ward to catch­ing up with the inter­na­tional UX com­mu­nity for the first time since mov­ing back to the U.S.

I’m shar­ing some of the work I’ve been doing at Endeca / Ora­cle, around the ques­tion of inter­ac­tion and sense mak­ing for the emerg­ing cat­e­gory of big data.  This is a UX ques­tion I don’t think is on the radar of many prac­ti­tion­ers.  And for those who are encoun­ter­ing it, the fram­ings I see for how to engage with this from the UX per­spec­tive are scat­tered and — frankly — small.  They tend to focus on the specifics of visu­al­iza­tion, and miss the larger pic­ture of how peo­ple engage in dis­cov­ery tasks and activ­i­ties every day, on small and large scales. If you’ve fol­lowed my work on other emerg­ing inter­ac­tion UX and inter­ac­tion spaces like enter­prise appli­ca­tions, games, mobile, social net­works, and — before I returned to my roots in mak­ing prod­ucts in a startup con­text — aug­mented real­ity, it’s easy to see I’m inter­ested in the deep struc­ture of new inter­ac­tion spaces, and I think a forward-looking per­spec­tive on the broad and fun­da­men­tal con­cep­tual frame of ref­er­ence for such new spaces is essen­tial for any­one who intends to work in them in a seri­ous and impact­ful fash­ion. So con­sider this talk an intro­duc­tion to the pack­age of ideas about tech­nol­ogy, inter­ac­tion, prod­ucts, and their dis­cov­ery aspects that I refer to as of the “Age of Insight” — the era in which every­one dis­cov­ers, and every­thing is discoverable.

The 2012 UXLX pro­gram is online, the talk is titled The Lan­guage of Dis­cov­ery: A Gram­mar for Design­ing Big Data Inter­ac­tions, and the ses­sion descrip­tion is below.

I hope to see a good mix of famil­iar and new faces at this grow­ing event.  Thanks to the orga­niz­ers for includ­ing me in the program!

The Lan­guage of Dis­cov­ery: A Gram­mar for Design­ing Big Data Interactions

The oncom­ing tidal wave of Big Data, with its rapidly evolv­ing ecosys­tem of multi-channel infor­ma­tion sat­u­rated envi­ron­ments and ser­vices, brings pro­found chal­lenges and oppor­tu­ni­ties for the design of effec­tive user experiences.

Look­ing deeper than the cel­e­bra­tory rhetoric of infor­ma­tion quan­tity, at its core, Big Data makes pos­si­ble unprece­dented aware­ness and insight into every sphere of life; from busi­ness and pol­i­tics, to the envi­ron­ment, arts and soci­ety. In this com­ing Age of Insight, ‘dis­cov­ery’ is not only the purview of spe­cial­ized Data Sci­en­tists who cre­ate exotic visu­al­iza­tions of mas­sive data sets, it is a fun­da­men­tal cat­e­gory of human activ­ity that is essen­tial to every­day inter­ac­tions between peo­ple, resources, and environments.

To pro­vide archi­tects and design­ers with an effec­tive start­ing point for cre­at­ing sat­is­fy­ing and rel­e­vant user expe­ri­ences that rely on dis­cov­ery inter­ac­tions, this ses­sion presents a sim­ple ana­lyt­i­cal and gen­er­a­tive toolkit for under­stand­ing how peo­ple con­duct the broad range of dis­cov­ery activ­i­ties nec­es­sary in the information-permeated world.

Specif­i­cally, this ses­sion will present:

• A sim­ple, research-derived lan­guage for describ­ing dis­cov­ery needs and activ­i­ties that spans domains, envi­ron­ments, media, and personas
• Observed and reusable pat­terns of dis­cov­ery activ­i­ties in indi­vid­ual and col­lab­o­ra­tive settings
• Exam­ples of the archi­tec­ture of suc­cess­ful dis­cov­ery expe­ri­ences at small and large scales
• A vocab­u­lary and per­spec­tive for dis­cov­ery as a crit­i­cal indi­vid­ual and orga­ni­za­tional capability
• Lead­ing edge exam­ples from the rapidly emerg­ing space of applied discovery
• Design futures and con­cepts explor­ing the pos­si­ble evo­lu­tion paths of dis­cov­ery interactions

The Euro­HCIR orga­niz­ers have pub­lished pro­ceed­ings from this year’s work­shop (which I was unfor­tu­nately unable to attend), which means I can make our paper A Tax­on­omy of Enter­prise Search and Dis­cov­ery directly avail­able.  The com­plete pro­ceed­ings are here, and are also pack­aged as a sin­gle down­load.

Here’s the text of the pub­lished paper, includ­ing ref­er­ences, and adding in a few illus­tra­tions omit­ted to meet page lim­its on papers.  Many thanks go to co-authors Tony Russell-Rose and Mark Bur­rell for putting this paper together.

A Tax­on­omy of Enter­prise Search

ABSTRACT

Clas­sic IR (infor­ma­tion retrieval) is pred­i­cated on the notion of users search­ing for infor­ma­tion in order to sat­isfy a par­tic­u­lar “infor­ma­tion need”. How­ever, it is now accepted that much of what we rec­og­nize as search behav­iour is often not infor­ma­tional per se. For exam­ple, Broder (2002) has shown that the need under­ly­ing a given web search could in fact be nav­i­ga­tional (e.g. to find a par­tic­u­lar site or known item) or trans­ac­tional (e.g. to find a sites through which the user can trans­act, e.g. through online shop­ping, social media, etc.). Sim­i­larly, Rose & Levin­son (2004) have iden­ti­fied con­sump­tion of online resources as a fur­ther cat­e­gory of search behav­iour and query intent.

In this paper, we extend this work to the enter­prise con­text, exam­in­ing the needs and behav­iours of indi­vid­u­als across a range of search and dis­cov­ery sce­nar­ios within var­i­ous types of enter­prise. We present an ini­tial tax­on­omy of “dis­cov­ery modes”, and dis­cuss some ini­tial impli­ca­tions for the design of more effec­tive search and dis­cov­ery plat­forms and tools.

Cat­e­gories and Sub­ject Descriptors

H.3.3 [Infor­ma­tion Search and Retrieval]: Search process;

H.3.5 [Online Infor­ma­tion Ser­vices]: Web-based services

Gen­eral Terms

Human Fac­tors.

Key­words

Enter­prise search, infor­ma­tion seek­ing, user behav­iour, knowl­edge work­ers, search modes, infor­ma­tion dis­cov­ery, user expe­ri­ence design.

1. INTRODUCTION

To design bet­ter search and dis­cov­ery expe­ri­ences we must under­stand the com­plex­i­ties of the human-information seek­ing process. Numer­ous the­o­ret­i­cal frame­works have been pro­posed to char­ac­ter­ize this com­plex process, notably the stan­dard model (Sut­cliffe & Ennis 1998), the cog­ni­tive model (Nor­man 1998) and the dynamic model (Bates, 1989). In addi­tion, oth­ers have inves­ti­gated search as a strate­gic process, exam­in­ing the var­i­ous prob­lem solv­ing strate­gies and tac­tics that infor­ma­tion seek­ers employ over extended peri­ods of time (e.g. Kuhlthau, 1991).

In this paper, we exam­ine the needs and behav­iours of var­ied indi­vid­u­als across a range of search and dis­cov­ery sce­nar­ios within var­i­ous types of enter­prise. These are based on an analy­sis of the sce­nar­ios derived from numer­ous engage­ments involv­ing the devel­op­ment of search and busi­ness intel­li­gence solu­tions uti­liz­ing the Endeca Lat­i­tude soft­ware plat­form. In so doing, we extend the clas­sic IR con­cept of information-seeking to a broader notion of discovery-oriented prob­lem solv­ing, accom­mo­dat­ing the much wider range of behav­iours required to ful­fil the typ­i­cal goals and objec­tives of enter­prise knowl­edge workers.

Our approach to enter­prise dis­cov­ery is an activity-centred model inspired by Don Norman’s Activ­ity Cen­tred Design, which “orga­nizes accord­ing to usage” whereas “…tra­di­tional human cen­tred design orga­nizes accord­ing to topic, in iso­la­tion, out­side the con­text of real, every­day use.” (Nor­man 2006). This approach is an exten­sion of pre­vi­ous activity-centred mod­el­ling efforts which focused on a “captur[ing] a sys­tem­atic and holis­tic view of what users need to accom­plish when under­tak­ing infor­ma­tion retrieval tasks more com­plex than search­ing” (Laman­tia 2006), employ­ing Grounded The­ory to pro­vide method­olog­i­cal struc­ture (Glaser 1967).

In this con­text, we present an alter­na­tive model focused on infor­ma­tion dis­cov­ery rather than infor­ma­tion seek­ing per se, which has at its core an ini­tial tax­on­omy of the “modes of dis­cov­ery” that knowl­edge work­ers employ to sat­isfy their infor­ma­tion search and dis­cov­ery goals. We then dis­cuss some ini­tial impli­ca­tions of this model for the design of more effec­tive search and dis­cov­ery plat­forms and tools.

2. INFORMATION RETRIEVAL MODELS

The clas­sic model of IR assumes an inter­ac­tion cycle con­sist­ing of four main activ­i­ties: the iden­ti­fi­ca­tion an infor­ma­tion need, the spec­i­fi­ca­tion of an appro­pri­ate query, the exam­i­na­tion of retrieval results, and refor­mu­la­tion (where nec­es­sary) of the orig­i­nal query. This cycle is then repeated until a suit­able result set is found (Salton 1989).

In both the above mod­els, the user’s infor­ma­tion need is assumed to be sta­tic. How­ever, it is now acknowl­edged that infor­ma­tion seek­ers’ needs often change as they inter­act with a search sys­tem. In recog­ni­tion of this, alter­na­tive mod­els of infor­ma­tion seek­ing have been pro­posed. For exam­ple, Bates (1989) pro­posed the dynamic “berry-picking” model of infor­ma­tion seek­ing, in which the infor­ma­tion need (and con­se­quently the query) changes through­out the search process This model also recog­nises that infor­ma­tion needs are not sat­is­fied by a sin­gle, final result set, but by the aggre­ga­tion of results, insights and inter­ac­tions along the way.

Bates’ work is par­tic­u­larly inter­est­ing as it explores the con­nec­tions between the dynamic model and the search strate­gies and tac­tics that pro­fes­sional information-seekers employ. In par­tic­u­lar, Bates iden­ti­fies a set of 29 indi­vid­ual tac­tics, organ­ised into four broad cat­e­gories (Bates, 1979). Like­wise, O’Day & Jef­fries (1993) exam­ined the use of infor­ma­tion search results by clients of pro­fes­sional infor­ma­tion inter­me­di­aries and iden­ti­fied three dis­tinct “search modes” or major cat­e­gories of search behav­iour: (1) Mon­i­tor­ing a known topic or set of vari­ables over time; (2) Fol­low­ing a spe­cific plan for infor­ma­tion gath­er­ing; (3) Explor­ing a topic in an undi­rected fashion.

O’Day and Jef­fries also observed that a given search would often evolve over time into a series of inter­con­nected searches, delim­ited by cer­tain trig­gers and stop con­di­tions that indi­cate the tran­si­tions between modes or indi­vid­ual searches exe­cuted as part of an over­all enquiry or sce­nario. More­over, O’Day & Jef­fries also attempted to char­ac­terise the analy­sis tech­niques employed by the clients in inter­pret­ing the search results, iden­ti­fy­ing the fol­low­ing six pri­mary cat­e­gories: (1) Look­ing for trends or cor­re­la­tions; (2) Mak­ing com­par­isons; (3) Exper­i­ment­ing with dif­fer­ent aggregations/scaling; (4) Iden­ti­fy­ing crit­i­cal sub­sets; (5) Mak­ing assess­ments; (6) Inter­pret­ing data to find meaning.

More recent inves­ti­ga­tions into the rela­tion­ship between infor­ma­tion needs and search activ­i­ties include that of Mar­chion­ini (2005), who iden­ti­fies three major cat­e­gories of search activ­ity, namely “Lookup”, “Learn” and “Investigate”.

3. A TAXONOMY OF ENTERPRISE SEACH AND DISCOVERY

The pri­mary source of data in this study is a set of user sce­nar­ios cap­tured dur­ing numer­ous engage­ments involv­ing the devel­op­ment of search and busi­ness intel­li­gence solu­tions uti­liz­ing the Endeca Lat­i­tude soft­ware plat­form. These sce­nar­ios take the form of a sim­ple nar­ra­tive that illus­trates the user’s end goal and the pri­mary task or action they take to com­plete it, fol­lowed by a brief descrip­tion of their job func­tion or role, for example:

“I need to under­stand a portfolio’s expo­sures to assess portfolio-level invest­ment mix” (Port­fo­lio Manager)

“I need to under­stand the qual­ity per­for­mance of a part and mod­ule set in man­u­fac­tur­ing and the field so that I can deter­mine if I should replace that part” (Engineering)

These sce­nar­ios were man­u­ally ana­lyzed to iden­tify themes or modes that appeared con­sis­tently through­out the set. For exam­ple, in each of the sce­nar­ios above there is an artic­u­la­tion of the need to develop an under­stand­ing or com­pre­hen­sion of some aspect of the data, imply­ing that “com­pre­hend­ing” may con­sti­tute one such dis­cov­ery mode. Inevitably, this analy­sis process was some­what iter­a­tive and sub­jec­tive, echo­ing the obser­va­tions made by Bates (1979) in the iden­ti­fi­ca­tion of her search tac­tics: “While our goal over the long term may be a par­si­mo­nious few, highly effec­tive tac­tics, our goal in the short term should be to uncover as many as we can, as being of poten­tial assis­tance. Then we can test the tac­tics and select the good ones. If we go for clo­sure too soon, i.e., seek that par­si­mo­nious few pre­ma­turely, then we may miss some valu­able tac­tics.”

There are how­ever some guid­ing prin­ci­ples that we can apply to facil­i­tate con­ver­gence on a sta­ble set. For exam­ple, an ideal set of modes would exhibit prop­er­ties such as: Con­sis­tency (they rep­re­sent approx­i­mately the same level of abstrac­tion); Orthog­o­nal­ity (they oper­ate inde­pen­dently to each other); and Com­pre­hen­sive­ness (they address the full range of dis­cov­ery scenarios).

The ini­tial set of dis­cov­ery modes to emerge from this analy­sis con­sists of a set of nine, arranged into three top-level cat­e­gories con­sis­tent with those of Mar­chion­ini (2005). The nine modes are as fol­lows, each shown with a brief definition:

1. Lookup

1a. Locat­ing: To find a spe­cific (pos­si­bly known) item;

1b. Ver­i­fy­ing: To con­firm or sub­stan­ti­ate that an item or set of items meets some spe­cific criterion;

1c. Mon­i­tor­ing: To main­tain aware­ness of the sta­tus of an item or data set for pur­poses of man­age­ment or control

2. Learn

2a. Com­par­ing: To exam­ine two or more items to iden­tify sim­i­lar­i­ties & differences;

2b. Com­pre­hend­ing: To gen­er­ate insight by under­stand­ing the nature or mean­ing of an item or data set;

2c. Explor­ing: To proac­tively inves­ti­gate or exam­ine an item or data set for the pur­pose of serendip­i­tous knowl­edge discovery

3. Inves­ti­gate

3a. Ana­lyz­ing: To crit­i­cally exam­ine the detail of an item or data set to iden­tify pat­terns & relationships;

3b. Eval­u­at­ing: To use judg­ment to deter­mine the sig­nif­i­cance or value of an item or data set with respect to a spe­cific bench­mark or model

3c. Syn­the­siz­ing: To gen­er­ate or com­mu­ni­cate insight by inte­grat­ing diverse inputs to cre­ate a novel arte­fact or com­pos­ite view

Evi­dently, the out­put of this process has been opti­mized for the cur­rent data set and in that respect rep­re­sents an ini­tial inter­pre­ta­tion that will need to evolve fur­ther. For exam­ple, “mon­i­tor­ing” may appear to be a lookup activ­ity when con­sid­ered in the con­text of a sim­ple alert mes­sage, but when viewed as a strate­gic activ­ity per­formed by an exec­u­tive in the con­text of an organ­i­sa­tional dash­board, a much greater degree of inter­ac­tion and com­plex­ity is implied. Con­versely, “explor­ing” is a con­cept whose level of abstrac­tion may prove some­what higher than the oth­ers, thus break­ing the con­sis­tency prin­ci­ple sug­gested above.

How­ever, the true value of the modes will be realised not by their con­cep­tual purity or ele­gance but by their util­ity as a design resource. In this respect, they should be judged by the extent to which they facil­i­tate the design process in cap­tur­ing impor­tant char­ac­ter­is­tics com­mon to enter­prise search and dis­cov­ery expe­ri­ences, whilst flex­i­bly accom­mo­dat­ing arbi­trary vari­a­tions in domain, infor­ma­tion resources, etc.

4. MODE SEQUENCES AND PATTERNS

A fur­ther inter­est­ing obser­va­tion aris­ing from the above analy­sis is that the map­ping between sce­nar­ios and modes is not one-to–one. Instead, some sce­nar­ios are seen to involve a num­ber of modes, some­times with a pri­mary or dom­i­nant mode, and often with an implied lin­ear sequence. More­over, cer­tain sequences of modes tend to re-occur more fre­quently than oth­ers, form­ing spe­cific “mode chains” or pat­terns, anal­o­gous to higher-level syn­tac­tic units. These pat­terns pro­vide a frame­work for under­stand­ing the tran­si­tions between modes (echo­ing the trig­gers iden­ti­fied by O’Day & Jef­fries), and allude to the exis­tence of nat­ural seams that can be used be used to pro­vide fur­ther insight into infor­ma­tion enter­prise search and dis­cov­ery behaviour.

These mode chains echo the above-mentioned efforts to cre­ate goal-based infor­ma­tion retrieval mod­els, which yielded modes and a set of broadly applic­a­ble “infor­ma­tion retrieval pat­terns that describe the ways users com­bine and switch modes to meet goals: Each pat­tern is assem­bled from com­bi­na­tions of the same four [ele­men­tal] modes” (Laman­tia 2006).

Fig­ure 1. Dis­cov­ery mode network

The five most fre­quent mode pat­terns are listed below. These have been assigned descrip­tive (if some­what infor­mal) labels to aid their char­ac­ter­i­sa­tion, along with the sequence of modes they rep­re­sent and an asso­ci­ated exam­ple scenario:

1. Comparison-driven opti­miza­tion: (Analyze-Compare– Eval­u­ate) e.g. “Replace a prob­lem­atic part with an equiv­a­lent or bet­ter part with­out com­pro­mis­ing qual­ity and cost”
2. Exploration-driven opti­miza­tion: (Explore-Analyze-Evaluate) e.g. “Iden­tify oppor­tu­ni­ties to opti­mize use of tool­ing capac­ity for my commodity/parts”
3. Strate­gic Insight (Analyze-Comprehend-Evaluate) e.g. “Under­stand a lead’s under­ly­ing posi­tions so that I can assess the qual­ity of the invest­ment oppor­tu­nity”
4. Strate­gic Over­sight (Monitor-Analyze-Evaluate) e.g. “Mon­i­tor & assess com­mod­ity sta­tus against strategy/plan/target”
5. Comparison-driven Syn­the­sis (Analyze-Compare-Synthesize) e.g. “Ana­lyze and under­stand consumer-customer-market trends to inform brand strat­egy & com­mu­ni­ca­tions plan”

Fur­ther insight may be derived by exam­in­ing how the mode pat­terns com­bine across all the sce­nar­ios to the form of a “mode net­work”, as shown in Fig­ure 1. Evi­dently, some modes act as “ter­mi­nal” nodes, i.e. entry points or exit points to a dis­cov­ery sce­nario. For exam­ple, Mon­i­tor and Explore fea­ture only as entry points at the ini­ti­a­tion of a sce­nario, whilst Syn­the­size and Eval­u­ate fea­ture only as exit points to a scenario.

5. DESIGN PRINCIPLES FOR SEARCH AND DISCOVERY SOLUTIONS

The modes estab­lish a ‘taskon­omy’ or col­lec­tion of defined dis­cov­ery activ­i­ties which are struc­turally con­sis­tent, domain and scale inde­pen­dent, orthog­o­nal, seman­ti­cally dis­tinct, con­cep­tu­ally con­nected, and flex­i­bly sequence­able. Such a pro­file — anal­o­gous to notes in the musi­cal scale, or the words and phrases we assem­ble into sen­tences — should allow the modes to serve as a lan­guage for the design of vari­able scale activity-centered dis­cov­ery solu­tions through com­mon con­struc­tive mech­a­nisms such as con­cate­na­tion, com­bi­na­tion and nest­ing. And if the modes do act as an ele­men­tary gram­mar for dis­cov­ery, then sus­tained use as a func­tional and inter­ac­tion design lan­guage should result in the cre­ation of larger and more com­plex units of mean­ing which offer cumu­la­tive value.

Pro­fes­sional expe­ri­ence with employ­ing the modes as both an ana­lyt­i­cal frame­work for under­stand­ing dis­cov­ery needs and as a design gram­mar for the def­i­n­i­tion of dis­cov­ery solu­tions sug­gests that both impli­ca­tions are valid. Fur­ther, our obser­va­tions of using the modes sug­gest the exis­tence of rec­og­niz­able pat­terns in the design of dis­cov­ery solu­tions. We will briefly dis­cuss some of the pat­terns observed, doing so at three com­mon lev­els of solu­tion scale: on the level of a sin­gle func­tional or inter­face ele­ment, for whole screens or inter­faces com­posed of mul­ti­ple func­tional ele­ments, and for appli­ca­tions com­pris­ing mul­ti­ple screens.

5.1 Sin­gle ele­ment patterns

5.1.1 Com­par­i­son Views

One of the most com­mon design pat­terns is to sup­port the need for the Com­pare mode by cre­at­ing A/B type com­par­i­son views that present two dis­play panes — each con­tain­ing data dis­play charts or tables; or sin­gle items or groups of items — side by side to empha­size sim­i­lar­i­ties and differences.

5.1.2 Con­tex­tual Views

Another com­mon design pat­tern sup­ports the Analy­sis mode by allow­ing a fore-grounded view of a sin­gle chart, table, item, or list, accom­pa­nied by its con­tex­tual ‘halo’ — the full body of infor­ma­tion avail­able about the ele­ment such as sta­tus, ori­gin, for­mat, rela­tion­ships to other ele­ments; anno­ta­tions; etc.

5.2 Whole screen patterns

5.2.1 Dash­board

One of the most com­mon screen-level design pat­terns is to sup­port the Mon­i­tor­ing and Syn­the­sis modes by pre­sent­ing a col­lec­tion of met­rics which in aggre­gate pro­vide the sta­tus of inde­pen­dent processes, groups, or progress ver­sus goals in a ‘dash­board’ style screen.

5.2.2 Visual Dis­cov­ery Screen: 4-Dimensions

 

A sec­ond com­mon screen-level design pat­tern for dis­cov­ery expe­ri­ences is the visual dis­cov­ery screen, which sup­ports modes such Explo­ration, Eval­u­a­tion, and Ver­i­fi­ca­tion by lay­er­ing views that present visu­al­iza­tions of sev­eral dimen­sions of a sin­gle axis of focus such as a core process, orga­ni­za­tional unit, or KPI. When switch­ing between lay­ered views, the axis in focus remains the same, but the data and pre­sen­ta­tion in the dimen­sions adjusts to match the pre­ferred dis­cov­ery mode.

5.3 Application-level patterns

5.3.1 Dif­fer­en­ti­ated Application

The ‘Dif­fer­en­ti­ated Appli­ca­tion’ pat­tern assem­bles a col­lec­tion of indi­vid­ual screens whose dis­tinct com­po­si­tions and designs sup­port indi­vid­ual dis­cov­ery modes of Analy­sis, Com­par­i­son, Eval­u­a­tion and Mon­i­tor­ing in aggre­gate to address the ‘Strate­gic Over­sight’ mode sequence. Application-level pat­terns often address a spec­trum of dis­cov­ery needs for a group of users with dif­fer­ing orga­ni­za­tional respon­si­bil­i­ties, such as man­age­ment vs. detailed analysis.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6. DISCUSSION

The above analy­sis is pred­i­cated on the notion that the user sce­nar­ios pro­vide a unique insight into the infor­ma­tion needs of enter­prise knowl­edge work­ers. How­ever, a num­ber of caveats apply to both the data and the approach.

Firstly, the sce­nar­ios were orig­i­nally gen­er­ated to sup­port the devel­op­ment of a spe­cific imple­men­ta­tion rather than for the analy­sis above. There­fore, the prin­ci­ples gov­ern­ing their cre­ation may not faith­fully reflect the true dis­tri­b­u­tion or pri­or­ity of infor­ma­tion needs among the var­i­ous end user pop­u­la­tions. Sec­ondly, the par­tic­u­lar sam­ple we selected for this study was based on a num­ber of prag­matic fac­tors (includ­ing avail­abil­ity), which may not faith­fully rep­re­sent the true dis­tri­b­u­tion or pri­or­ity among enter­prise orga­ni­za­tions. Thirdly, the data will inevitably con­tain some degree of sub­jec­tiv­ity, par­tic­u­larly in cases where sce­nar­ios were gen­er­ated by proxy rather than with direct end-user con­tact. Fourthly, the data will inevitably con­tain some degree of incon­sis­tency in cases where sce­nar­ios were doc­u­mented by dif­fer­ent individuals.

We should also acknowl­edge a num­ber of caveats con­cern­ing the process itself. In induc­tive work with foun­da­tions in qual­i­ta­tively cen­tered frame­works such as Grounded The­ory, it is expected that a num­ber of iter­a­tions of a “propose-classify-refine” cycle will be required for the process to con­verge on a sta­ble out­put (e.g. Rose & Levin­son, 2004). In addi­tion, those iter­a­tions should involve a vari­ety of crit­i­cal view­points, with the out­put tested and refined using a sep­a­rate, inde­pen­dent sam­ple on each iter­a­tion. Like­wise, the process by which sce­nar­ios are clas­si­fied would ben­e­fit from fur­ther rigour: this is a crit­i­cal part of the process and of course relies on human judge­ment and infer­ence, but that judge­ment needs to go beyond sim­ple word match­ing and be con­sis­tently applied to each sce­nario so that sub­tle dis­tinc­tions in mean­ing and intent can be accu­rately iden­ti­fied and recorded.

That said, some inter­est­ing com­par­isons can already be made with the exist­ing frame­works. For exam­ple, the first and third of the search modes sug­gested by O’Day and Jef­fries have also been iden­ti­fied as dis­tinct dis­cov­ery modes in our own study, and the sec­ond (arguably) maps on to one or more of the mode chains iden­ti­fied above. Like­wise, the search results analy­sis tech­niques that O’Day & Jef­fries iden­ti­fied also present some inter­est­ing parallels.

7. CONCLUSIONS AND FUTURE DIRECTIONS

To design bet­ter search and dis­cov­ery expe­ri­ences we must under­stand the com­plex­i­ties of the human-information seek­ing process. In this paper, we have exam­ined the needs and behav­iours of var­ied indi­vid­u­als across a range of search and dis­cov­ery sce­nar­ios within var­i­ous types of enter­prise. In so doing, we have extended the clas­sic IR con­cept of information-seeking to a broader notion of discovery-oriented prob­lem solv­ing, accom­mo­dat­ing the much wider range of behav­iours required to ful­fil the typ­i­cal goals and objec­tives of enter­prise knowl­edge workers.

In addi­tion, we have pro­posed an alter­na­tive model focused on infor­ma­tion dis­cov­ery rather than infor­ma­tion seek­ing which has at its core a tax­on­omy of “modes of dis­cov­ery” that knowl­edge work­ers employ to sat­isfy their infor­ma­tion search and dis­cov­ery goals. We have also exam­ined some of the ini­tial impli­ca­tions of this model for the design of more effec­tive search and dis­cov­ery plat­forms and tools.

Sug­ges­tions for future work include fur­ther iter­a­tions on the “propose-classify-refine” cycle using inde­pen­dent data. This data should ide­ally be acquired based on a prin­ci­pled sam­pling strat­egy that attempts where pos­si­ble to address any biases intro­duced in the cre­ation of the orig­i­nal sce­nar­ios. In addi­tion, this process should be com­ple­mented by empir­i­cal research and obser­va­tion of knowl­edge work­ers in con­text to val­i­date and refine the dis­cov­ery modes and trig­gers that give rise to the observed pat­terns of usage.

8. REFERENCES

[1] Bates, Mar­cia J. 1979. “Infor­ma­tion Search Tac­tics.” Jour­nal of the Amer­i­can Soci­ety for Infor­ma­tion Sci­ence 30: 205–214

[2] Bates, Mar­cia J. 1989. “The Design of Brows­ing and Berryp­ick­ing Tech­niques for the Online Search Inter­face.” Online Review 13: 407–424.

[3] Broder, A. 2002. A tax­on­omy of web search, ACM SIGIR Forum, v.36 n.2, Fall 2002

[4] Kuhlthau, C. C. 1991. Inside the infor­ma­tion search process: Infor­ma­tion seek­ing from the user’s per­spec­tive. Jour­nal of the Amer­i­can Soci­ety for Infor­ma­tion Sci­ence, 42, 361–371.

[5] Laman­tia, J. 2006. “10 Infor­ma­tion Retrieval Pat­terns” JoeLamantia.com, http://www.joelamantia.com/information-architecture/10-information-retrieval-patterns

[6] Glaser, B. & Strauss, A. 1967. The Dis­cov­ery of Grounded The­ory: Strate­gies for Qual­i­ta­tive Research. New York: Aldine de Gruyter.

[7] Mar­chion­ini, G. 2006. Exploratory search: from find­ing to under­stand­ing. Com­mun. ACM 49(4): 41–46

[8] Nor­man, Don­ald A. 1988. The psy­chol­ogy of every­day things. New York, NY, US: Basic Books.

[9] Don­ald A. Nor­man. 2006. Logic ver­sus usage: the case for activ­ity cen­tered design. Inter­ac­tions 13, 6

[10] O’Day, V. and Jef­fries, R. 1993. Ori­en­teer­ing in an infor­ma­tion land­scape: how infor­ma­tion seek­ers get from here to there. INTERCHI 1993: 438–445

[11] Rose, D. and Levin­son, D. 2004. Under­stand­ing user goals in web search, Pro­ceed­ings of the 13^th inter­na­tional con­fer­ence on World Wide Web, New York, NY, USA

[12] Salton, G. (1989). Auto­matic Text Pro­cess­ing: The Trans­for­ma­tion, Analy­sis, and Retrieval of Infor­ma­tion by Com­puter. Addison-Wesley, Read­ing, MA.

[13] A.G. Sut­cliffe and M. Ennis. Towards a cog­ni­tive the­ory of infor­ma­tion retrieval. Inter­act­ing with Com­put­ers, 10:321–351, 1998.

 

 

I’m pleased to be pre­sent­ing ‘A Tax­on­omy of Enter­prise Search’ at the upcom­ing Euro­HCIR work­shop, part of the 2011 HCI con­fer­ence in the UK.  Co-authored with Tony Russell-Rose of UXLabs, and Mark Bur­rell here at Endeca, this is our first pub­li­ca­tion of some of the very excit­ing work we’re doing to under­stand and describe dis­cov­ery activ­i­ties in enter­prise set­tings, and do so within a changed and broader fram­ing than tra­di­tional infor­ma­tion retrieval.  The paper builds on work I’ve done pre­vi­ously on under­stand­ing and defin­ing infor­ma­tion needs and pat­terns of infor­ma­tion retrieval activ­ity, while work­ing on search and dis­cov­ery prob­lems as part of larger user expe­ri­ence archi­tec­ture efforts.

Here’s the abstract of the paper:

Clas­sic IR (infor­ma­tion retrieval) is pred­i­cated on the notion of users search­ing for infor­ma­tion in order to sat­isfy a par­tic­u­lar “infor­ma­tion need”. How­ever, it is now accepted that much of what we rec­og­nize as search behav­iour is often not infor­ma­tional per se. For exam­ple, Broder (2002) has shown that the need under­ly­ing a given web search could in fact be nav­i­ga­tional (e.g. to find a par­tic­u­lar site or known item) or trans­ac­tional (e.g. to find a sites through which the user can trans­act, e.g. through online shop­ping, social media, etc.). Sim­i­larly, Rose & Levin­son (2004) have iden­ti­fied con­sump­tion of online resources as a fur­ther cat­e­gory of search behav­iour and query intent.

In this paper, we extend this work to the enter­prise con­text, exam­in­ing the needs and behav­iours of indi­vid­u­als across a range of search and dis­cov­ery sce­nar­ios within var­i­ous types of enter­prise. We present an ini­tial tax­on­omy of “dis­cov­ery modes”, and dis­cuss some ini­tial impli­ca­tions for the design of more effec­tive search and dis­cov­ery plat­forms and tools.

There’s a con­sid­er­able amount of research avail­able on infor­ma­tion retrieval — even within a com­par­a­tively new dis­ci­pline like HCIR, focused on the human to sys­tem inter­ac­tion aspects of IR — but I think it’s the attempt to define an activ­ity cen­tered gram­mar for inter­act­ing with infor­ma­tion that makes our approach worth exam­in­ing.  The HCIR events in the U.S. (and now Europe) blend aca­d­e­mic and prac­ti­tioner per­spec­tives, so are an appro­pri­ate audi­ence for our pro­posed vocab­u­lary of dis­cov­ery activ­ity ‘modes’ that’s based on a sub­stan­tial body of data col­lected and ana­lyzed dur­ing solu­tion design and deploy­ment engagements.

I’ll post the paper itself once the pro­ceed­ings are available.

 

 

 

Endeca invites cus­tomers, part­ners and lead­ing mem­bers of the broader search and dis­cov­ery tech­nol­ogy and solu­tions com­mu­ni­ties to meet annu­ally, and show­case the most inter­est­ing and excit­ing work in the field of dis­cov­ery.  As lead for the UX team that designs Endeca’s dis­cov­ery prod­ucts, I shared some of our recent work on pat­terns in the struc­ture of dis­cov­ery appli­ca­tions, as well as best prac­tices in infor­ma­tion design and visu­al­iza­tion that we use to drive prod­uct def­i­n­i­tion and design for Endeca’s Lat­i­tude Dis­cov­ery Framework.

This mate­r­ial is use­ful for pro­gram and project man­agers and busi­ness ana­lysts defin­ing require­ments for dis­cov­ery solu­tions and appli­ca­tions, UX and sys­tem archi­tects craft­ing high-level struc­tures and address­ing long-term growth, inter­ac­tion design­ers and tech­ni­cal devel­op­ers defin­ing and build­ing infor­ma­tion work­spaces at a fine grain, and

There are three major sec­tions: the first presents some of our tools for iden­ti­fy­ing and under­stand­ing people’s needs and goals for dis­cov­ery in terms of activ­ity (the Lan­guage of Dis­cov­ery as we call it), the sec­ond brings together screen-level, appli­ca­tion level, and user sce­nario / use-case level pat­terns we’ve observed in the appli­ca­tions cre­ated to meet those needs, and the final sec­tion shares con­densed best prac­tices and fun­da­men­tal prin­ci­ples for infor­ma­tion design and visu­al­iza­tion based on aca­d­e­mic research dis­ci­plines such as cog­ni­tive sci­ence and infor­ma­tion retrieval.

It’s no coin­ci­dence that these sec­tions reflect the appli­ca­tion of the core UX dis­ci­plines of user research, infor­ma­tion archi­tec­ture, and inter­ac­tion design to the ques­tion of “who will need to encounter infor­ma­tion for some end, and in what kind of expe­ri­ence will they encounter it”.  This flow and order­ing is delib­er­ate; it demon­strates on two lev­els the results of our own efforts apply­ing the UX per­spec­tive to the ques­tions inher­ent in cre­at­ing dis­cov­ery tools, and shares some of the tools, insights, tem­plates, and resources we use to shape the plat­form used to cre­ate dis­cov­ery expe­ri­ences across diverse industries.

Ses­sion outline

1. Under­stand­ing User Needs
2. Design Pat­terns for Dis­cov­ery Applications
3. Design Prin­ci­ples and Gudielines for Infor­ma­tion Inter­ac­tion and Visualization

Ses­sion description

“How can you har­ness the power and flex­i­bil­ity of Lat­i­tude to cre­ate use­ful, usable, and com­pelling dis­cov­ery appli­ca­tions for enter­prise dis­cov­ery work­ers? This ses­sion goes beyond the tech­nol­ogy to explore how you can apply fun­da­men­tal prin­ci­ples of infor­ma­tion design and visu­al­iza­tion, ana­lyt­ics best prac­tices and user inter­face design pat­terns to com­pose effec­tive and com­pelling dis­cov­ery appli­ca­tions that opti­mize user dis­cov­ery, suc­cess, engage­ment, & adoption.”

The pat­terns are prod­uct spe­cific in that they show how to com­pose screens and appli­ca­tions using the pre­de­fined com­po­nents in the Dis­cov­ery Frame­work library.  How­ever, many of the product-specific com­po­nents are built to address com­mon or recur­ring needs for inter­ac­tion with infor­ma­tion via well-known mech­a­nisms such as search, fil­ter­ing, nav­i­ga­tion, visu­al­iza­tion, and pre­sen­ta­tion of data.  In other words, even if you’re not using the lit­eral Dis­cov­ery Frame­work com­po­nent library to com­pose your spe­cific infor­ma­tion analy­sis work­space, you’ll find these pat­terns rel­e­vant at work­space and appli­ca­tion lev­els of scale.

The deeper story of these pat­terns is in demon­strat­ing the evo­lu­tion of dis­cov­ery and analy­sis appli­ca­tions over time.  Typ­i­cally, dis­cov­ery appli­ca­tions begin by offer­ing users a general-purpose work­space that sat­is­fies a wide range of inter­ac­tion tasks in an approx­i­mate fash­ion.  Over time, via suc­ces­sive expan­sions in the the scope and vari­ety of data they present, and the dis­cov­ery and analy­sis capa­bil­i­ties they pro­vide, dis­cov­ery appli­ca­tions grow to include sev­eral dif­fer­ent types of work­spaces that indi­vid­u­ally address dis­tinct sets of needs for visu­al­iza­tion and sense mak­ing by using very dif­fer­ent com­bi­na­tions of com­po­nents.  As a com­pos­ite, these func­tional and infor­ma­tion­ally diverse work­spaces span the full range of inter­ac­tion needs for dif­fer­ing types of users.

I hope you find this toolkit and col­lec­tion of pat­terns and infor­ma­tion design prin­ci­ples use­ful.  What are some of the resources you’re using to take on these challenges?

The good peo­ple at Chi Nether­lands just posted video of my talk “Play­ing Well With Oth­ers: Inter­ac­tion Design and Social Design for Aug­mented Real­ity” at the Web and Beyond 2010 here in Ams­ter­dam in June.  It’s couched as a col­lec­tion of design prin­ci­ples for the oncom­ing cat­e­gory of social aug­mented inter­ac­tions made pos­si­ble by the new medium of aug­mented real­ity.  But this talk is also a call to action for all mak­ers of expe­ri­ences for the emerg­ing engage­ment space of every­ware to focus on the human and the humane per­spec­tives as we explore the new inter­ac­tions made possible.

The out­line of the talk is roughly:

1. Overview of aug­mented reality
2. Social inter­ac­tion per­spec­tive on cur­rent AR experiences
3. Def­i­n­i­tion of ‘social aug­mented experiences’
4. Com­mon inter­ac­tion design pat­terns for AR
5. Social ‘anti-patterns’ lim­it­ing design of aug­mented experiences
6. Design prin­ci­ples for social aug­mented experiences

(The audio qual­ity is quite good, and the cam­era­man cap­tured most of the slides nicely — so this is a record­ing worth watching.)

This year’s TWAB fea­tured sev­eral talks on aug­mented real­ity, ubiq­ui­tous com­put­ing and related top­ics; you’ll find record­ings of these on the Chi Ned­er­land Vimeo chan­nel: http://vimeo.com/chinederland

Many thanks to the orga­niz­ers and vol­un­teers for putting on such a well-run event!

TWAB2010: Joe Laman­tia — Play­ing well with oth­ers: inter­ac­tion design and social design for aug­mented real­ity from Chi Ned­er­land on Vimeo.

Thanks to all who came to the Muziekge­bouw on a lovely early sum­mer day to talk about the emerg­ing engage­ment space of social aug­mented expe­ri­ences for the third edi­tion of The Web And Beyond con­fer­ence in Amsterdam.

For ref­er­ence, here’s the ses­sion descrip­tion from the offi­cial program:

Aug­mented real­ity blends the real world and the Inter­net in real time, mak­ing many new kinds of prox­im­ity, con­text, and loca­tion based expe­ri­ences pos­si­ble for indi­vid­u­als and groups. Despite these many pos­si­bil­i­ties, we know from his­tory that the long term value and impact of aug­mented real­ity for most peo­ple will depend on how well these expe­ri­ences inte­grate with ordi­nary social set­tings, and sup­port every­day inter­ac­tions. Yet the inter­ac­tion pat­terns and behav­ior we see in cur­rent AR expe­ri­ences seem almost ‘anti-social’ by design. This is an impor­tant gap that design must close in order to cre­ate suc­cess­ful AR offer­ings. In other words, much like chil­dren going to school for the first time, AR must to learn to ‘play well with oth­ers’ to be valu­able and suc­cess­ful. This pre­sen­ta­tion reviews the inter­ac­tion design pat­terns com­mon to aug­mented real­ity, sug­gests tools to help under­stand and improve the ’social matu­rity’ of AR prod­ucts and appli­ca­tions, and shares design prin­ci­ples for cre­at­ing gen­uinely social aug­mented expe­ri­ences that inte­grate well with human social set­tings and interactions.

I’m post­ing slides for my ‘Under­stand­ing Frame­works’ por­tion of the Beyond Find­abil­ity work­shop on strate­gic prac­tices just given at the 2010 IA Sum­mit.  This por­tion of the full-day pro­gram empha­sizes under­stand­ing and iden­ti­fy­ing the com­mon things that make up a design frame­work, con­cen­trat­ing on the sim­ple struc­ture that design­ers need to grasp in order to cre­ate their own effec­tive frame­works for solv­ing design chal­lenges. I hope you find it infor­ma­tive and useful!

I’ve posted my slides for the Where 2.0 panel “The Next Wave of AR: Social Aug­mented Expe­ri­ences” orga­nized by Tish Shute. After a review of the cur­rent state of aug­mented real­ity expe­ri­ences in terms of the social inter­ac­tions sup­ported (using the met­ric of ‘social matu­rity’), it shares 9 prin­ci­ples for cre­at­ing social AR expe­ri­ences that peo­ple will enjoy and value.

Spe­cial points to those who spot the embed­ded April Fool’s joke…